Cumulus Networks

Cumulus Linux 4.0 User Guide

What is Cumulus Linux?

Cumulus Linux is the first full-featured Linux operating system for the networking industry. The Debian Buster-based, networking-focused distribution runs on hardware produced by a broad partner ecosystem, ensuring unmatched customer choice regarding silicon, optics, cables, and systems.

This user guide provides in-depth documentation on the Cumulus Linux installation process, system configuration and management, network solutions, and monitoring and troubleshooting recommendations. In addition, the quick start guide provides an end-to-end setup process to get you started.

Cumulus Linux 4.0 includes the latest NetQ agent and CLI, which is installed by default on the Cumulus Linux switch. Use NetQ to monitor and manage your data center network infrastructure and operational health. Refer to the NetQ documentation for details.

For information on new features, bug fixes, and known issues present in this release, refer to the release notes.

Open Source Contributions

To implement various Cumulus Linux features, Cumulus Networks has forked various software projects, like CFEngine Netdev and some Puppet Labs packages. Some of the forked code resides in the Cumulus Networks GitHub repository and some is available as part of the Cumulus Linux repository as Debian source packages.

Hardware Compatibility List

You can find the most up-to-date hardware compatibility list (HCL) here. Use the HCL to confirm that your switch model is supported by Cumulus Networks. The HCL is updated regularly, listing products by port configuration, manufacturer and SKU part number.

Stay up to Date

Quick Start Guide

This quick start guide provides an end-to-end setup process for installing and running Cumulus Linux, as well as a collection of example commands for getting started after installation is complete.

Prerequisites

Intermediate-level Linux knowledge is assumed for this guide. You need to be familiar with basic text editing, Unix file permissions, and process monitoring. A variety of text editors are pre-installed, including vi and nano.

You must have access to a Linux or UNIX shell. If you are running Windows, use a Linux environment like Cygwin as your command line tool for interacting with Cumulus Linux.

If you are a networking engineer but are unfamiliar with Linux concepts, refer to this reference guide to compare the Cumulus Linux CLI and configuration options, and their equivalent Cisco Nexus 3000 NX-OS commands and settings. You can also watch a series of short videos introducing you to Linux and Cumulus Linux-specific concepts.

Install Cumulus Linux

To install Cumulus Linux, you use ONIE (Open Network Install Environment), an extension to the traditional U-Boot software that allows for automatic discovery of a network installer image. This facilitates the ecosystem model of procuring switches with an operating system choice, such as Cumulus Linux. The easiest way to install Cumulus Linux with ONIE is with local HTTP discovery:

  1. If your host (laptop or server) is IPv6-enabled, make sure it is running a web server. If the host is IPv4-enabled, make sure it is running DHCP in addition to a web server.

  2. Download the Cumulus Linux installation file to the root directory of the web server. Rename this file onie-installer.

  3. Connect your host using an Ethernet cable to the management Ethernet port of the switch.

  4. Power on the switch. The switch downloads the ONIE image installer and boots. You can watch the progress of the install in your terminal. After the installation completes, the Cumulus Linux login prompt appears in the terminal window.

These steps describe a flexible unattended installation method. You do not need a console cable. A fresh install with ONIE using a local web server typically completes in less than ten minutes.

You have more options for installing Cumulus Linux with ONIE. Read Installing a New Cumulus Linux Image to install Cumulus Linux using ONIE in the following ways:

  • DHCP/web server with and without DHCP options
  • Web server without DHCP
  • FTP or TFTP without a web server
  • Local file
  • USB

After installing Cumulus Linux, you are ready to:

Getting Started

When starting Cumulus Linux for the first time, the management port makes a DHCPv4 request. To determine the IP address of the switch, you can cross reference the MAC address of the switch with your DHCP server. The MAC address is typically located on the side of the switch or on the box in which the unit ships.

Login Credentials

The default installation includes the system account (root), with full system privileges and the user account (cumulus), with sudo privileges. The root account password is locked by default (which prohibits login). The cumulus account is configured with this default password:

CumulusLinux!

In this quick start guide, you use the cumulus account to configure Cumulus Linux.

For optimum security, change the default password with the passwd command before you configure Cumulus Linux on the switch.

All accounts except root are permitted remote SSH login; you can use sudo to grant a non-root account root-level access. Commands that change the system configuration require this elevated level of access.

For more information about sudo, read Using sudo to Delegate Privileges.

Serial Console Management

You are encouraged to perform management and configuration over the network, either in band or out of band. A serial console is fully supported; however, you might prefer the convenience of network-based management.

Typically, switches ship from the manufacturer with a mating DB9 serial cable. Switches with ONIE are always set to a 115200 baud rate.

Wired Ethernet Management

Switches supported in Cumulus Linux always contain at least one dedicated Ethernet management port, which is named eth0. This interface is geared specifically for out-of-band management use. The management interface uses DHCPv4 for addressing by default. You can set a static IP address with the Network Command Line Utility (NCLU) or by editing the /etc/network/interfaces file (Linux).

NCLU Commands

Set the static IP address with the interface address and interface gateway NCLU commands:

cumulus@switch:~$ net add interface eth0 ip address 192.0.2.42/24
cumulus@switch:~$ net add interface eth0 ip gateway 192.0.2.1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Set a static IP address by editing the /etc/network/interfaces file:

cumulus@switch:~$ sudo nano /etc/network/interfaces
# Management interface
auto eth0
iface eth0
    address 192.0.2.42/24
    gateway 192.0.2.1

Configure the Hostname and Timezone

Configure the hostname and timezone for your switch. The hostname identifies the switch; make sure you configure the hostname to be unique and descriptive.

  • Do not use an underscore (_) in the hostname; underscores are not permitted.
  • Avoid using apostrophes or non-ASCII characters in the hostname. Cumulus Linux does not parse these characters.

To change the hostname:

NCLU Commands

Run the net add hostname command, which modifies both the /etc/hostname and /etc/hosts files with the desired hostname.

cumulus@switch:~$ net add hostname <hostname>
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

  • The command prompt in the terminal does not reflect the new hostname until you either log out of the switch or start a new shell.
  • When you use the NCLU command to set the hostname, DHCP does not override the hostname when you reboot the switch. However, if you disable the hostname setting with NCLU, DHCP does override the hostname the next time you reboot the switch.

Linux Commands
  1. Modify the /etc/hostname file with the desired hostname:
cumulus@switch:~$ sudo nano /etc/hostname
  1. In /etc/hosts file, replace the 127.0.1.1 IP address with the new hostname:
cumulus@switch:~$ sudo nano /etc/hosts

The default timezone on the switch is (Coordinated Universal Time) UTC. Change the timezone on your switch to be the timezone for your location.

To update the timezone, use NTP interactive mode:

  1. Run the following command in a terminal.
cumulus@switch:~$ sudo dpkg-reconfigure tzdata
  1. Follow the on screen menu options to select the geographic area and region.

Programs that are already running (including log files) and users currently logged in, do not see timezone changes made with interactive mode. To set the timezone for all services and daemons, reboot the switch.

Verify the System Time

Before you install the license, verify that the date and time on the switch are correct, and correct the date and time if necessary. If the date and time is incorrect, the switch might not be able to synchronize with Puppet or might return errors after you restart switchd:

Warning: Unit file of switchd.service changed on disk, 'systemctl daemon-reload' recommended.

Install the License

Cumulus Linux is licensed on a per-instance basis. Each network system is fully operational, enabling any capability to be utilized on the switch with the exception of forwarding on switch panel ports. Only eth0 and console ports are activated on an unlicensed instance of Cumulus Linux. Enabling front panel ports requires a license.

You receive a license key from Cumulus Networks or an authorized reseller. Here is a sample license key:

user@company.com|thequickbrownfoxjumpsoverthelazydog312

There are three ways to install the license onto the switch:

cumulus@switch:~$ scp user@my_server:/home/user/my_license_file.txt .
cumulus@switch:~$ sudo cl-license -i my_license_file.txt
cumulus@switch:~$ sudo cl-license -i <URL>
cumulus@switch:~$ sudo cl-license -i
<paste license key>
^+d

Check that your license is installed with the cl-license command.

cumulus@switch:~$ cl-license 
user@example.com|$ampleL1cen$et3xt

It is not necessary to reboot the switch to activate the switch ports. After you install the license, restart the switchd service. All front panel ports become active and show up as swp1, swp2, and so on.

cumulus@switch:~$ sudo systemctl restart switchd.service

If a license is not installed on a Cumulus Linux switch, the switchd service does not start. After you install the license, start switchd as described above.

Configure Breakout Ports with Splitter Cables

If you are using 4x10G DAC or AOC cables, or want to break out 100G or 40G switch ports, configure the breakout ports. For more details, see Switch Port Attributes.

Test Cable Connectivity

By default, all data plane ports (every Ethernet port except the management interface, eth0) are disabled.

To test cable connectivity:

NCLU Commands

To administratively enable a port:

cumulus@switch:~$ net add interface swp1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To administratively enable all physical ports, run the following command, where swp1-52 represents a switch with switch ports numbered from swp1 to swp52:

cumulus@switch:~$ net add interface swp1-52
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To view link status, use the net show interface all command. The following examples show the output of ports in admin down, down, and up modes:

cumulus@switch:~$ net show interface all
State  Name           Spd  MTU    Mode           LLDP                    Summary
-----  -------------  ---  -----  -------------  ----------------------  -------------------------
UP     lo             N/A  65536  Loopback                               IP: 127.0.0.1/8
       lo                                                                IP: 10.0.0.11/32
       lo                                                                IP: 10.0.0.112/32
       lo                                                                IP: ::1/128
UP     eth0           1G   1500   Mgmt           oob-mgmt-switch (swp6)  Master: mgmt(UP)
       eth0                                                              IP: 192.168.0.11/24(DHCP)
UP     swp1           1G   9000   BondMember     server01 (eth1)         Master: bond01(UP)
UP     swp2           1G   9000   BondMember     server02 (eth1)         Master: bond02(UP)
ADMDN  swp45          N/A  1500   NotConfigured
ADMDN  swp46          N/A  1500   NotConfigured
ADMDN  swp47          N/A  1500   NotConfigured
ADMDN  swp48          N/A  1500   NotConfigured
UP     swp49          1G   9000   BondMember     leaf02 (swp49)          Master: peerlink(UP)
UP     swp50          1G   9000   BondMember     leaf02 (swp50)          Master: peerlink(UP)
UP     swp51          1G   9216   NotConfigured  spine01 (swp1)
UP     swp52          1G   9216   NotConfigured  spine02 (swp1)
UP     bond01         1G   9000   802.3ad                                Master: bridge(UP)
       bond01                                                            Bond Members: swp1(UP)
UP     bond02         1G   9000   802.3ad                                Master: bridge(UP)
       bond02                                                            Bond Members: swp2(UP)
UP     bridge         N/A  1500   Bridge/L2
UP     mgmt           N/A  65536  Interface/L3                           IP: 127.0.0.1/8
UP     peerlink       2G   9000   802.3ad                                Master: bridge(UP)
       peerlink                                                          Bond Members: swp49(UP)
       peerlink                                                          Bond Members: swp50(UP)
DN     peerlink.4094  2G   9000   SubInt/L3                              IP: 169.254.1.1/30
ADMDN  vagrant        N/A  1500   NotConfigured
UP     vlan13         N/A  1500   Interface/L3                           Master: vrf1(UP)
       vlan13                                                            IP: 10.1.3.11/24
UP     vlan13-v0      N/A  1500   Interface/L3                           Master: vrf1(UP)
       vlan13-v0                                                         IP: 10.1.3.1/24
UP     vlan24         N/A  1500   Interface/L3                           Master: vrf1(UP)
       vlan24                                                            IP: 10.2.4.11/24
UP     vlan24-v0      N/A  1500   Interface/L3                           Master: vrf1(UP)
       vlan24-v0                                                         IP: 10.2.4.1/24
UP     vlan4001       N/A  1500   NotConfigured                          Master: vrf1(UP)
UP     vni13          N/A  9000   Access/L2                              Master: bridge(UP)
UP     vni24          N/A  9000   Access/L2                              Master: bridge(UP)
UP     vrf1           N/A  65536  NotConfigured
UP     vxlan4001      N/A  1500   Access/L2                              Master: bridge(UP)
Linux Commands

To enable a port, run the ip link set <interface> up command. For example:

cumulus@switch:~$ sudo ip link set swp1 up

As root, run the following bash script to administratively enable all physical ports:

cumulus@switch:~$ sudo su -
cumulus@switch:~$ for i in /sys/class/net/*; do iface=`basename $i`; if [[ $iface == swp* ]]; then ip link set $iface up fi done

To view link status, use the ip link show command. The following examples show the output of a port in down and up mode:

# Administratively Down
swp1: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT qlen 1000

# Administratively Up but Layer 1 protocol is Down
swp1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT qlen 500

# Administratively Up, Layer 1 protocol is Up
swp1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT qlen 500

Configure Switch Ports

Layer 2 Port Configuration

Cumulus Linux does not put all ports into a bridge by default. To create a bridge and configure one or more front panel ports as members of the bridge, use the following examples as a guide.

NCLU Commands

In the following configuration example, the front panel port swp1 is placed into a bridge called bridge.

cumulus@switch:~$ net add bridge bridge ports swp1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

You can add a range of ports in one command. For example, to add swp1 through swp10, swp12, and swp14 through swp20 to bridge:

cumulus@switch:~$ net add bridge bridge ports swp1-10,12,14-20
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

In the following configuration example, the front panel port swp1 is placed into a bridge called br0:

    ...
    auto br0
    iface br0
      bridge-ports swp1
      bridge-stp on

To put a range of ports into a bridge, use the glob keyword. For example, to add swp1 through swp10, swp12, and swp14 through swp20 to br0:

    ...
    auto br0
    iface br0
      bridge-ports glob swp1-10 swp12 glob swp14-20
      bridge-stp on

To activate or apply the configuration to the kernel:

# First, check for typos:
cumulus@switch:~$ sudo ifquery -a

# Then activate the change if no errors are found:
cumulus@switch:~$ sudo ifup -a

To view the changes in the kernel, use the brctl command:

cumulus@switch:~$ brctl show
bridge name     bridge id              STP enabled     interfaces
br0             8000.089e01cedcc2       yes              swp1

Layer 3 Port Configuration

You can also configure a front panel port or bridge interface as a layer 3 port.

NCLU Commands

In the following configuration example, the front panel port swp1 is configured as a layer 3 access port:

cumulus@switch:~$ net add interface swp1 ip address 10.1.1.1/30
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To add an IP address to a bridge interface, you must put it into a VLAN interface. If you want to use a VLAN other than the native one, set the bridge PVID:

cumulus@switch:~$ net add vlan 100 ip address 10.2.2.1/24
cumulus@switch:~$ net add bridge bridge pvid 100
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

In the following configuration example, the front panel port swp1 is configured as a layer 3 access port:

auto swp1
iface swp1
  address 10.1.1.1/30

To add an IP address to a bridge interface, include the address under the iface stanza in the /etc/network/interfaces file. If you want to use a VLAN other than the native one, set the bridge PVID:

auto br0
iface br0
    address 10.2.2.1/24
    bridge-ports glob swp1-10 swp12 glob swp14-20
    bridge-pvid 100
    bridge-stp on

To activate or apply the configuration to the kernel:

# First check for typos:
cumulus@switch:~$ sudo ifquery -a

# Then activate the change if no errors are found:
cumulus@switch:~$ sudo ifup -a

To view the changes in the kernel, use the ip addr show command:

cumulus@switch:~$ ip addr show
...
4. swp1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bridge state UP group default qlen 1000
        link/ether 44:38:39:00:6e:fe brd ff:ff:ff:ff:ff:ff
...
14: bridge: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default 
    link/ether 44:38:39:00:00:04 brd ff:ff:ff:ff:ff:ff
    inet6 fe80::4638:39ff:fe00:4/64 scope link 
        valid_lft forever preferred_lft forever
...

Configure a Loopback Interface

Cumulus Linux has a loopback preconfigured in the /etc/network/interfaces file. When the switch boots up, it has a loopback interface, called lo, which is up and assigned an IP address of 127.0.0.1.

The loopback interface lo must always be specified in the /etc/network/interfaces file and must always be up.

To see the status of the loopback interface (lo):

NCLU Commands

Use the net show interface lo command.

cumulus@switch:~$ net show interface lo
    Name    MAC                Speed    MTU    Mode
--  ------  -----------------  -------  -----  --------
UP  lo      00:00:00:00:00:00  N/A      65536  Loopback

Alias
-----
loopback interface
IP Details
-------------------------  --------------------
IP:                        127.0.0.1/8, ::1/128
IP Neighbor(ARP) Entries:  0

The loopback is up and is assigned an IP address of 127.0.0.1.

To add an IP address to a loopback interface, configure the lo interface:

cumulus@switch:~$ net add loopback lo ip address 10.1.1.1/32
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

You can configure multiple loopback addresses by adding additional address lines:

cumulus@switch:~$ net add loopback lo ip address 172.16.2.1/24
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Use the ip addr show lo command.

cumulus@switch:~$ ip addr show lo
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
        valid_lft forever preferred_lft forever

The loopback is up and is assigned an IP address of 127.0.0.1.

To add an IP address to a loopback interface, add it directly under the iface lo inet loopback definition in the /etc network/interfaces file:

auto lo
iface lo inet loopback
    address 10.1.1.1

If an IP address is configured without a mask (as shown above), the IP address becomes a /32. So, in the above case, 10.1.1.1 is actually 10.1.1.1/32.

You can add multiple loopback addresses by adding additional address lines in the /etc/network/interfaces file:

auto lo
iface lo inet loopback
    address 10.1.1.1
    address 172.16.2.1/24

Reboot the Switch

After you complete the configuration in this section, reboot the switch:

cumulus@switch:~$ sudo reboot

Installation Management

This section describes how to manage, install, and upgrade Cumulus Linux on your switch.

Managing Cumulus Linux Disk Images

The Cumulus Linux operating system resides on a switch as a disk image. This section discusses how to manage the disk image.

To install a new Cumulus Linux disk image, refer to Installing a New Cumulus Linux Image. To upgrade Cumulus Linux, refer to Upgrading Cumulus Linux.

Determine the Switch Platform

To determine if your switch is on an x86 or ARM platform, run the uname -m command.

For example, on an x86 platform, uname -m outputs x86_64:

cumulus@switch:~$ uname -m
 x86_64

On an ARM platform, uname -m outputs armv7l:

cumulus@switch:~$ uname -m
 armv7l

You can also visit the HCL (hardware compatibility list) to look at your hardware and determine the processor type.

Reprovision the System (Restart the Installer)

Reprovisioning the system deletes all system data from the switch.

To stage an ONIE installer from the network (where ONIE automatically locates the installer), run the onie-select -i command. A reboot is required for the reinstall to begin.

cumulus@switch:~$ sudo onie-select -i
WARNING:
WARNING: Operating System install requested.
WARNING: This will wipe out all system data.
WARNING:
Are you sure (y/N)? y
Enabling install at next reboot...done.
Reboot required to take effect.

To cancel a pending reinstall operation, run the onie-select -c command:

cumulus@switch:~$ sudo onie-select -c
Cancelling pending install at next reboot...done.

To stage an installer located in a specific location, run the onie-install -i command. You can specify a local, absolute or relative path, an HTTP or HTTPS server, SCP or FTP server. You can also stage a Zero Touch Provisioning (ZTP) script along with the installer. The onie-install command is typically used with the -a option to activate installation. If you do not specify the -a option, a reboot is required for the reinstall to begin.

The following example stages the installer located at http://203.0.113.10/image-installer together with the ZTP script located at http://203.0.113.10/ztp-script and activates installation and ZTP:

cumulus@switch:~$ sudo onie-install -i http://203.0.113.10/image-installer
cumulus@switch:~$ sudo onie-install -z http://203.0.113.10/ztp-script
cumulus@switch:~$ sudo onie-install -a

You can also specify these options together in the same command. For example:

cumulus@switch:~$ sudo onie-install -i http://203.0.113.10/image-installer -z http://203.0.113.10/ztp-script -a

To see more onie-install options, run man onie-install.

Uninstall All Images and Remove the Configuration

To remove all installed images and configurations, and return the switch to its factory defaults, run the onie-select -k command.

The onie-select -k command takes a long time to run as it overwrites the entire NOS section of the flash. Only use this command if you want to erase all NOS data and take the switch out of service.

cumulus@switch:~$ sudo onie-select -k
WARNING:
WARNING: Operating System uninstall requested.
WARNING: This will wipe out all system data.
WARNING:
Are you sure (y/N)? y
Enabling uninstall at next reboot...done.
Reboot required to take effect.

A reboot is required for the uninstallation process to begin.

To cancel a pending uninstall operation, run the onie-select -c command:

cumulus@switch:~$ sudo onie-select -c
Cancelling pending uninstall at next reboot...done.

Boot into Rescue Mode

If your system becomes unresponsive is some way, you can correct certain issues by booting into ONIE rescue mode. In rescue mode, the file systems are unmounted and you can use various Cumulus Linux utilities to try and resolve a problem.

To reboot the system into ONIE rescue mode, run the onie-select -r command:

cumulus@switch:~$ sudo onie-select -r
WARNING:
WARNING: Rescue boot requested.
WARNING:
Are you sure (y/N)? y
Enabling rescue at next reboot...done.
Reboot required to take effect.

A reboot is required to boot into rescue mode.

To cancel a pending rescue boot operation, run the onie-select -c command:

cumulus@switch:~$ sudo onie-select -c
Cancelling pending rescue at next reboot...done.

Inspect the Image File

The Cumulus Linux installation disk image file is executable. From a running switch, you can display, extract, and verify the contents of the image file.

To display the contents of the Cumulus Linux image file, pass the info option to the image file. For example, to display the contents of an image file called onie-installer located in the /var/lib/cumulus/installer directory:

cumulus@switch:~$ sudo /var/lib/cumulus/installer/onie-installer info
Verifying image checksum ...OK.
Preparing image archive ... OK.
Control File Contents
=====================
Description: Cumulus Linux 4.0.0
Release: 4.0.0
Architecture: amd64
Switch-Architecture: bcm-amd64
Build-Id: dirtyz224615f
Build-Date: 2019-05-17T16:34:22+00:00
Build-User: clbuilder
Homepage: http://www.cumulusnetworks.com/
Min-Disk-Size: 1073741824
Min-Ram-Size: 536870912
mkimage-version: 0.11.111_gbcf0

To extract the contents of the image file, use with the extract <path> option. For example, to extract an image file called onie-installer located in the /var/lib/cumulus/installer directory to the mypath directory:

cumulus@switch:~$ sudo /var/lib/cumulus/installer/onie-installer extract mypath
total 181860
-rw-r--r-- 1 4000 4000       308 May 16 19:04 control
drwxr-xr-x 5 4000 4000      4096 Apr 26 21:28 embedded-installer
-rw-r--r-- 1 4000 4000  13273936 May 16 19:04 initrd
-rw-r--r-- 1 4000 4000   4239088 May 16 19:04 kernel
-rw-r--r-- 1 4000 4000 168701528 May 16 19:04 sysroot.tar

To verify the contents of the image file, use with the verify option. For example, to verify the contents of an image file called onie-installer located in the /var/lib/cumulus/installer directory:

cumulus@switch:~$ sudo /var/lib/cumulus/installer/onie-installer verify
Verifying image checksum ...OK.
Preparing image archive ... OK.
./cumulus-linux-bcm-amd64.bin.1: 161: ./cumulus-linux-bcm-amd64.bin.1: onie-sysinfo: not found
Verifying image compatibility ...OK.
Verifying system ram ...OK.
Open Network Install Environment (ONIE) Home Page

Installing a New Cumulus Linux Image

You can install a new Cumulus Linux disk image using ONIE, an open source project (equivalent to PXE on servers) that enables the installation of network operating systems (NOS) on bare metal switches.

Before you install Cumulus Linux, the switch can be in two different states:

The sections below describe some of the different ways you can install the Cumulus Linux disk image, such as using a DHCP/web server, FTP, a local file, or a USB drive. Steps are provided for both installing directly from ONIE (if no image is installed on the switch) and from Cumulus Linux (if the image is already installed on the switch), where applicable. For additional methods to find and install the Cumulus Linux image, see the ONIE Design Specification.

You can download a Cumulus Linux image from the Cumulus Networks Downloads page.

Installing the Cumulus Linux disk image is destructive; configuration files on the switch are not saved; copy them to a different server before installing.

In the following procedures:

Install Using a DHCP/Web Server with DHCP Options

To install Cumulus Linux using a DHCP/web server with DHCP options, set up a DHCP/web server on your laptop and connect the eth0 management port of the switch to your laptop. After you connect the cable, the installation proceeds as follows:

  1. The bare metal switch boots up and requests an IP address (DHCP request).
  2. The DHCP server acknowledges and responds with DHCP option 114 and the location of the installation image.
  3. ONIE downloads the Cumulus Linux disk image, installs, and reboots.
  4. Success! You are now running Cumulus Linux.

The most common method is to send DHCP option 114 with the entire URL to the web server (this can be the same system). However, there are many other ways to use DHCP even if you do not have full control over DHCP. See the ONIE user guide for help with partial installer URLs and advanced DHCP options; both articles list more supported DHCP options.

Here is an example DHCP configuration with an ISC DHCP server:

subnet 172.0.24.0 netmask 255.255.255.0 {
  range 172.0.24.20 172.0.24.200;
  option default-url = "http://172.0.24.14/onie-installer-[PLATFORM]";
}

Here is an example DHCP configuration with dnsmasq (static address assignment):

dhcp-host=sw4,192.168.100.14,6c:64:1a:00:03:ba,set:sw4
dhcp-option=tag:sw4,114,"http://roz.rtplab.test/onie-installer-[PLATFORM]"

If you do not have a web server, you can use this free Apache example.

Install Using a DHCP/Web Server without DHCP Options

Follow the steps below if you can log into the switch on a serial console (ONIE), or log in on the console or with ssh (Install from Cumulus Linux).

Install from ONIE
  1. Place the Cumulus Linux disk image in a directory on the web server.
  2. Run the onie-nos-install command:
ONIE:/ #onie-nos-install http://10.0.1.251/path/to/cumulus-install-[PLATFORM].bin
Install from Cumulus Linux
  1. Place the Cumulus Linux disk image in a directory on the web server.

  2. From the Cumulus Linux command prompt, run the onie-install command, then reboot the switch.

    cumulus@switch:~$ sudo onie-install -a -i http://10.0.1.251/path/to/cumulus-install-[PLATFORM].bin
    

Install Using a Web Server with no DHCP

Follow the steps below if you can log into the switch on a serial console (ONIE), or log in on the console or with ssh (Install from Cumulus Linux) but no DHCP server is available.

You need a console connection to access the switch; you cannot perform this procedure remotely.

Install from ONIE
  1. ONIE is in discovery mode. You must disable discovery mode with the following command:

    onie# onie-discovery-stop
    

    On older ONIE versions, if the onie-discovery-stop command is not supported, run:

    onie# /etc/init.d/discover.sh stop
    
  2. Assign a static address to eth0 with the ip addr add command:

    ONIE:/ #ip addr add 10.0.1.252/24 dev eth0
    
  3. Place the Cumulus Linux disk image in a directory on your web server.

  4. Run the installer manually (because there are no DHCP options):

    ONIE:/ #onie-nos-install http://10.0.1.251/path/to/cumulus-install-[PLATFORM].bin
    
Install from Cumulus Linux
  1. Place the Cumulus Linux disk image in a directory on your web server.

  2. From the Cumulus Linux command prompt, run the onie-install command, then reboot the switch.

    cumulus@switch:~$ sudo onie-install -a -i http://10.0.1.251/path/to/cumulus-install-[PLATFORM].bin
    

Install Using FTP Without a Web Server

Follow the steps below if your laptop is on the same network as the switch eth0 interface but no DHCP server is available.

Install from ONIE
  1. Set up DHCP or static addressing for eth0. The following example assigns a static address to eth0:

    ONIE:/ #ip addr add 10.0.1.252/24 dev eth0
    
  2. If you are using static addressing, disable ONIE discovery mode:

    onie# onie-discovery-stop
    

    On older ONIE versions, if the onie-discovery-stop command is not supported, run:

    onie# /etc/init.d/discover.sh stop
    
  3. Place the Cumulus Linux disk image into a TFTP or FTP directory.

  4. If you are not using DHCP options, run one of the following commands (tftp for TFTP or ftp for FTP):

    ONIE# onie-nos-install ftp://local-ftp-server/cumulus-install-[PLATFORM].bin
    
    ONIE# onie-nos-install tftp://local-tftp-server/cumulus-install-[PLATFORM].bin
    
Install from Cumulus Linux
  1. Place the Cumulus Linux disk image into a TFTP or FTP directory.

  2. From the Cumulus Linux command prompt, run one of the following commands (tftp for TFTP or ftp for FTP), then reboot the switch.

    cumulus@switch:~$ sudo onie-install -a -i ftp://local-ftp-server/cumulus-install-[PLATFORM].bin
    
    cumulus@switch:~$ sudo onie-install -a -i tftp://local-ftp-server/cumulus-install-[PLATFORM].bin
    

Install Using a Local File

Follow the steps below to install the disk image referencing a local file.

Install from ONIE
  1. Set up DHCP or static addressing for eth0. The following example assigns a static address to eth0:

    ONIE:/ #ip addr add 10.0.1.252/24 dev eth0
    
  2. If you are using static addressing, disable ONIE discovery mode.

    onie# onie-discovery-stop
    

    On older ONIE versions, if the onie-discovery-stop command is not supported, run:

    onie# /etc/init.d/discover.sh stop
    
  3. Use scp to copy the Cumulus Linux disk image to the switch.

  4. Run the installer manually from ONIE:

    ONIE:/ #onie-nos-install /path/to/local/file/cumulus-install-[PLATFORM].bin
    
Install from Cumulus Linux
  1. Copy the Cumulus Linux disk image to the switch.

  2. From the Cumulus Linux command prompt, run the onie-install command, then reboot the switch.

    cumulus@switch:~$ sudo onie-install -a -i /path/to/local/file/cumulus-install-[PLATFORM].bin
    

Install Using a USB Drive

Follow the steps below to install the Cumulus Linux disk image using a USB drive. Instructions are provided for x86 and ARM platforms.

  • Installing Cumulus Linux using a USB drive is fine for a single switch here and there but is not scalable. DHCP can scale to hundreds of switch installs with zero manual input unlike USB installs.
  • Cumulus Networks also provides Cumulus on a Stick, which packages Cumulus Linux images with your license. You can download your personalized ZIP file, transfer it to a USB drive, insert the drive into your switch, apply power, and you are ready to go. See Cumulus on a Stick for information.

Prepare for USB Installation

  1. From the Cumulus Networks Downloads page, download the appropriate Cumulus Linux image for your x86 or ARM platform.

  2. From a computer, prepare your USB drive by formatting it using one of the supported formats: FAT32, vFAT or EXT2.

    Optional: Prepare a USB Drive inside Cumulus Linux

    Use caution when performing the actions below; it is possible to severely damage your system with the following utilities.
    

    1. Insert your USB drive into the USB port on the switch running Cumulus Linux and log in to the switch. Examine output from cat /proc/partitions and sudo fdisk -l [device] to determine on which device your USB drive can be found. For example, sudo fdisk -l /dev/sdb.

      These instructions assume your USB drive is the /dev/sdb device, which is typical if you insert the USB drive after the machine is already booted. However, if you insert the USB drive during the boot process, it is possible that your USB drive is the /dev/sda device. Make sure to modify the commands below to use the proper device for your USB drive.

    2. Create a new partition table on the USB drive. (The parted utility should already be installed. However, if it is not, install it with sudo -E apt-get install parted.)

    sudo parted /dev/sdb mklabel msdos
    
    1. Create a new partition on the USB drive:
    sudo parted /dev/sdb -a optimal mkpart primary 0% 100%
    
    1. Format the partition to your filesystem of choice using one of the examples below:
    sudo mkfs.ext2 /dev/sdb1
    sudo mkfs.msdos -F 32 /dev/sdb1
    sudo mkfs.vfat /dev/sdb1
    

    To use mkfs.msdos or mkfs.vfat, you need to install the dosfstools package from the Debian software repositories, as they are not included by default.

    1. To continue installing Cumulus Linux, mount the USB drive to move files
    sudo mkdir /mnt/usb
    sudo mount /dev/sdb1 /mnt/usb
    
  3. Copy the Cumulus Linux disk image to the USB drive, then rename the image file to:

    • onie-installer-x86_64, if installing on an x86 platform
    • onie-installer-arm, if installing on an ARM platform

    You can also use any of the ONIE naming schemes mentioned here.

    When using a Mac or Windows computer to rename the installation file, the file extension might still be present. Make sure to remove the file extension otherwise ONIE is not able to detect the file.

  4. Insert the USB drive into the switch, then continue with the appropriate instructions below for your x86 or ARM platform.

Instructions for x86 Platforms

Click to expand x86 instructions...
  1. Prepare the switch for installation:

    • If the switch is offline, connect to the console and power on the switch.
    • If the switch is already online in ONIE, use the reboot command.

    SSH sessions to the switch get dropped after this step. To complete the remaining instructions, connect to the console of the switch. Cumulus Linux switches display their boot process to the console; you need to monitor the console specifically to complete the next step.

  2. Monitor the console and select the ONIE option from the first GRUB screen shown below.

  3. Cumulus Linux on x86 uses GRUB chainloading to present a second GRUB menu specific to the ONIE partition. No action is necessary in this menu to select the default option ONIE: Install OS.

  4. The USB drive is recognized and mounted automatically. The image file is located and automatic installation of Cumulus Linux begins. Here is some sample output:

ONIE: OS Install Mode  ...

Version : quanta_common_rangeley-2014.05.05-6919d98-201410171013
Build  Date: 2014-10-17T10:13+0800
Info: Mounting kernel filesystems...  done.
Info: Mounting LABEL=ONIE-BOOT on /mnt/onie-boot  ...
initializing eth0...
scsi 6:0:0:0: Direct-Access  SanDisk Cruzer Facet 1.26 PQ: 0 ANSI: 6
sd 6:0:0:0: [sdb] 31266816 512-byte logical blocks: (16.0 GB/14.9 GiB)
sd 6:0:0:0: [sdb] Write Protect is off
sd 6:0:0:0: [sdb] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sd 6:0:0:0: [sdb] Attached SCSI disk

<...snip...>

ONIE:  Executing installer: file://dev/sdb1/onie-installer-x86_64
Verifying image checksum ... OK.
Preparing image archive ... OK.
Dumping image info...
Control File Contents
=====================
Description: Cumulus Linux
OS-Release:  4.0.0~1571178373.763fd151
Architecture: amd64
Date:  Fri, 22 November 2019 17:10:30 -0700
Installer-Version:  1.2
Platforms: accton_as5712_54x accton_as6712_32x  mlx_sx1400_i73612 dell_s4000_c2338 dell_s3000_c2338  cel_redstone_xp cel_smallstone_xp cel_pebble quanta_panther  quanta_ly8_rangeley quanta_ly6_rangeley quanta_ly9_rangeley 

Homepage: http://www.cumulusnetworks.com/
  1. After installation completes, the switch automatically reboots into the newly installed instance of Cumulus Linux.

Instructions for ARM Platforms

Click to expand ARM instructions...
  1. Prepare the switch for installation:

    • If the switch is offline, connect to the console and power on the switch.
    • If the switch is already online in ONIE, use the reboot command.

    SSH sessions to the switch get dropped after this step. To complete the remaining instructions, connect to the console of the switch. Cumulus Linux switches display their boot process to the console; you need to monitor the console specifically to complete the next step.

  2. Interrupt the normal boot process before the countdown (shown below) completes. Press any key to stop the autoboot.

U-Boot 2013.01-00016-gddbf4a9-dirty (Feb 14 2014 - 16:30:46) Accton: 1.4.0.5

CPU0: P2020, Version: 2.1, (0x80e20021)
Core: E500, Version: 5.1, (0x80211051)
Clock Configuration:
      CPU0:1200 MHz, CPU1:1200 MHz, 
      CCB:600 MHz,
      DDR:400 MHz (800 MT/s data rate) (Asynchronous), LBC:37.500 MHz
L1: D-cache 32 kB enabled
I-cache 32 kB enabled

<...snip ...>

USB: USB2513 hub OK
Hit any key to stop autoboot: 0
  1. A command prompt appears so that you can run commands. Execute the following command:
run onie_bootcmd
  1. The USB drive is recognized and mounted automatically. The image file is located and automatic installation of Cumulus Linux begins. Here is some sample output:
Loading Open Network Install Environment  ...
Platform: arm-as4610_54p-r0
Version : 1.6.1.3
WARNING: adjusting available memory to 30000000
## Booting kernel from Legacy Image at ec040000  ...
       Image Name:   as6701_32x.1.6.1.3
       Image Type:   ARM Linux Multi-File Image (gzip compressed)
       Data Size:    4456555 Bytes = 4.3 MiB
       Load Address: 00000000
       Entry Point:  00000000
       Contents:
          Image 0: 3738543 Bytes = 3.6 MiB
          Image 1: 706440 Bytes = 689.9 KiB
          Image 2: 11555 Bytes = 11.3 KiB
   Verifying Checksum ... OK
## Loading init Ramdisk from multi component Legacy Image at ec040000  ...
## Flattened Device Tree from multi component Image at EC040000
   Booting using the fdt at 0xec47d388
   Uncompressing Multi-File Image ... OK
   Loading Ramdisk to 2ff53000, end 2ffff788 ... OK
   Loading Device Tree to 03ffa000, end 03fffd22 ... OK

<...snip...>

ONIE: Starting ONIE Service Discovery
ONIE: Executing installer: file://dev/sdb1/onie-installer-arm
Verifying image checksum ... OK.
Preparing image archive ... OK.
Dumping image info ...
Control File Contents
=====================
Description: Cumulus Linux
OS-Release: 3.0.0-3b46bef-201509041633-build
Architecture: arm
Date: Fri, 27 May 2016 17:08:35 -0700
Installer-Version: 1.2
Platforms: accton_as4600_54t, accton_as6701_32x, accton_5652, accton_as5610_52x, dni_6448, dni_7448, dni_c7448n, cel_kennisis, cel_redstone, cel_smallstone, cumulus_p2020, quanta_lb9, quanta_ly2, quanta_ly2r, quanta_ly6_p2020
Homepage: http://www.cumulusnetworks.com/
  1. After installation completes, the switch automatically reboots into the newly installed instance of Cumulus Linux.

Upgrading Cumulus Linux

This topic describes how to upgrade Cumulus Linux on your switch.

Cumulus Networks recommends that you deploy, provision, configure, and upgrade switches using automation, even with small networks or test labs. During the upgrade process, you can quickly upgrade dozens of devices in a repeatable manner. Using tools like Ansible, Chef, or Puppet for configuration management greatly increases the speed and accuracy of the next major upgrade; these tools also enable the quick swap of failed switch hardware.

Before You Upgrade

Be sure to read the knowledge base article Upgrades: Network Device and Linux Host Worldview Comparison, which provides a detailed comparison between the network device and Linux host worldview of upgrade and installation.

Understanding the location of configuration data is required for successful upgrades, migrations, and backup. As with other Linux distributions, the /etc directory is the primary location for all configuration data in Cumulus Linux. The following list is a likely set of files that you need to back up and migrate to a new release. Make sure you examine any file that has been changed. Cumulus Networks recommends you consider making the following files and directories part of a backup strategy.

Network Configuration Files
File Name and LocationExplanationCumulus Linux DocumentationDebian Documentation
/etc/network/Network configuration files, most notably /etc/network/interfaces and /etc/network/interfaces.d/Switch Port AttributesN/A
/etc/resolv.confDNS resolutionNot unique to Cumulus Linux: wiki.debian.org/NetworkConfigurationhttps://www.debian.org/doc/manuals/debian-reference/ch05.en.html
/etc/frr/Routing application (responsible for BGP and OSPF)FRRouting OverviewN/A
/etc/hostnameConfiguration file for the hostname of the switchQuick Start Guidehttps://wiki.debian.org/HowTo/ChangeHostname
/etc/hostsConfiguration file for the hostname of the switchQuick Start Guidehttps://wiki.debian.org/HowTo/ChangeHostname
/etc/cumulus/acl/*Netfilter configurationNetfilter - ACLsN/A
/etc/cumulus/ports.confBreakout cable configuration fileSwitch Port AttributesN/A; read the guide on breakout cables
/etc/cumulus/switchd.confswitchd configurationConfiguring switchdN/A; read the guide on switchd configuration
Additional Commonly Used Files
File Name and LocationExplanationCumulus Linux DocumentationDebian Documentation
/etc/motdMessage of the dayNot unique to Cumulus Linuxwiki.debian.org/motd
/etc/passwdUser account informationNot unique to Cumulus Linuxhttps://www.debian.org/doc/manuals/debian-reference/ch04.en.html
/etc/shadowSecure user account informationNot unique to Cumulus Linuxhttps://www.debian.org/doc/manuals/debian-reference/ch04.en.html
/etc/groupDefines user groups on the switchNot unique to Cumulus Linuxhttps://www.debian.org/doc/manuals/debian-reference/ch04.en.html
/etc/lldpd.confLink Layer Discover Protocol (LLDP) daemon configurationLink Layer Discovery Protocolhttps://packages.debian.org/buster/lldpd
/etc/lldpd.d/Configuration directory for lldpdLink Layer Discovery Protocolhttps://packages.debian.org/buster/lldpd
/etc/nsswitch.confName Service Switch (NSS) configuration fileTACACS+N/A
/etc/ssh/SSH configuration filesSSH for Remote Accesshttps://wiki.debian.org/SSH
/etc/sudoers, /etc/sudoers.dBest practice is to place changes in /etc/sudoers.d/ instead of /etc/sudoers; changes in the /etc/sudoers.d/ directory are not lost during upgrade.Using sudo to Delegate Privileges

  • If you are using the root user account, consider including /root/.
  • If you have custom user accounts, consider including /home/<username>/.
  • Cumulus Networks recommends you run the net show configuration files | grep -B 1 "===" command and back up the files listed in the command output.

Never Migrate these Files between Versions or Switches
File Name and LocationExplanation
/etc/bcm.d/Per-platform hardware configuration directory, created on first boot. Do not copy.
/etc/mlx/Per-platform hardware configuration directory, created on first boot. Do not copy.
/etc/default/clagdCreated and managed by ifupdown2. Do not copy.
/etc/default/grubGrub init table. Do not modify manually.
/etc/default/hwclockPlatform hardware-specific file. Created during first boot. Do not copy.
/etc/initPlatform initialization files. Do not copy.
/etc/init.d/Platform initialization files. Do not copy.
/etc/fstabStatic information on filesystem. Do not copy.
/etc/image-releaseSystem version data. Do not copy.
/etc/os-releaseSystem version data. Do not copy.
/etc/lsb-releaseSystem version data. Do not copy.
/etc/lvm/archiveFilesystem files. Do not copy.
/etc/lvm/backupFilesystem files. Do not copy.
/etc/modulesCreated during first boot. Do not copy.
/etc/modules-load.d/Created during first boot. Do not copy.
/etc/sensors.dPlatform-specific sensor data. Created during first boot. Do not copy.
/root/.ansibleAnsible tmp files. Do not copy.
/home/cumulus/.ansibleAnsible tmp files. Do not copy.

If you are using certain forms of network virtualization, including VMware NSX-V or Midokura MidoNet, you might have updated the /usr/share/openvswitch/scripts/ovs-ctl-vtep file. This file is not marked as a configuration file; therefore, if the file contents change in a newer release of Cumulus Linux, they overwrite any changes you made to the file. Be sure to back up this file and the database file conf.db before upgrading.

You can check which files have changed since the last binary install with the following commands. Be sure to back up any changed files:

  • Run the sudo dpkg --verify command to show a list of changed files.
  • Run the egrep -v '^$|^#|=""$' /etc/default/isc-dhcp-* command to see if any of the generated /etc/default/isc-* files have changed.

Upgrade Cumulus Linux

To upgrade to Cumulus Linux 4.0 from Cumulus Linux 3.7, you must install a disk image of the new release using ONIE. You cannot upgrade packages with the apt-get upgrade command.

ONIE is an open source project (equivalent to PXE on servers) that enables the installation of network operating systems (NOS) on a bare metal switch.

Upgrading an MLAG pair requires additional steps. If you are using MLAG to dual connect two Cumulus Linux switches in your environment, follow the steps in Upgrading Cumulus Linux below to ensure a smooth upgrade.

Cumulus Networks deprecated lightweight network virtualization (LNV) in Cumulus Linux 4.0 in favor of Ethernet virtual private networks (EVPN. If your network is configured for LNV, you need to migrate your network configuration to a BGP EVPN configuration that is functionally equivalent before you upgrade to Cumulus Linux 4.0. Refer to Migrating from LNV to EVPN..

Be aware of the following when installing the disk image:

To upgrade the switch:

  1. Back up the configurations off the switch.
  2. Download the Cumulus Linux image.
  3. Install the disk image with the onie-install -a -i <image-location> command, which boots the switch into ONIE. The following example command installs the image from a web server, then reboots the switch. There are additional ways to install the disk image, such as using FTP, a local file, or a USB drive. For more information, see Installing a New Cumulus Linux Image.
cumulus@switch:~$ sudo onie-install -a -i http://10.0.1.251/cumulus-linux-4.0.0-mlx-amd64.bin && sudo reboot
  1. Restore the configuration files to the new release - ideally with automation.
  2. Verify correct operation with the old configurations on the new release.
  3. Reinstall third party applications and associated configurations.

Upgrade Switches in an MLAG Pair

If you are using MLAG to dual connect two switches in your environment, follow the steps below to upgrade the switches.

For networks with MLAG deployments, Cumulus Networks only supports upgrading to Cumulus Linux 4.0 from version 3.7.10 or later. If you are using a version of Cumulus Linux earlier than 3.7.10, you must upgrade to version 3.7.10 first, then upgrade to version 4.0. Version 3.7.10 is available on the downloads page on our website.

  1. Verify the switch is in the secondary role:
cumulus@switch:~$ clagctl status
  1. Shut down the core uplink layer 3 interfaces:
cumulus@switch:~$ sudo ip link set swpX down
  1. Shut down the peer link:
cumulus@switch:~$ sudo ip link set peerlink down
  1. Run the onie-install -a -i <image-location> command to boot the switch into ONIE. The following example command installs the image from a web server. There are additional ways to install the disk image, such as using FTP, a local file, or a USB drive. For more information, see Installing a New Cumulus Linux Image.
cumulus@switch:~$ sudo onie-install -a -i http://10.0.1.251/downloads/cumulus-linux-4.0.0-mlx-amd64.bin
  1. Reboot the switch:
cumulus@switch:~$ sudo reboot
  1. Verify STP convergence across both switches:
cumulus@switch:~$ mstpctl showall
  1. Verify core uplinks and peer links are UP:
cumulus@switch:~$ net show interface
  1. Verify MLAG convergence:
cumulus@switch:~$ clagctl status
  1. Make this secondary switch the primary:
cumulus@switch:~$ clagctl priority 2048
  1. Verify the other switch is now in the secondary role.
  2. Repeat steps 2-8 on the new secondary switch.
  3. Remove the priority 2048 and restore the priority back to 32768 on the current primary switch:
cumulus@switch:~$ clagctl priority 32768

Roll Back a Cumulus Linux Installation

Even the most well planned and tested upgrades can result in unforeseen problems; sometimes the best solution is to roll back to the previous state. There are three main strategies; all require detailed planning and execution:

The method you employ is specific to your deployment strategy, so providing detailed steps for each scenario is outside the scope of this document.

Third Party Packages

Third party packages in the Linux host world often use the same package system as the distribution into which it is to be installed (for example, Debian uses apt-get). Or, the package might be compiled and installed by the system administrator. Configuration and executable files generally follow the same filesystem hierarchy standards as other applications.

If you install any third party applications on a Cumulus Linux switch, configuration data is typically installed into the /etc directory, but it is not guaranteed. It is your responsibility to understand the behavior and configuration file information of any third party packages installed on the switch.

After you upgrade using a full disk image install, you need to reinstall any third party packages or any Cumulus Linux add-on packages.

Migrating from LNV to EVPN

Cumulus Networks deprecated lightweight network virtualization (LNV) in Cumulus Linux 4.0 in favor of Ethernet virtual private networks (EVPN) to enable interoperability with switches from other manufacturers, to commit to industry standards, and because the benefits of EVPN outweigh those of LNV.

If your network is configured for LNV, you need to migrate your network configuration to a BGP EVPN configuration that is functionally equivalent before you upgrade to Cumulus Linux 4.0 or later.

Migration Considerations

You cannot run LNV and EVPN at the same time for the following reasons:

Upgrade to EVPN

Cumulus Networks highly recommends that you use automation, such as Ansible to upgrade to EVPN. Automation ensures minimal downtime, reduces human error, and is useful at almost any scale.

Cumulus Networks also recommends you use NCLU to update the configuration for the following reasons:

The upgrade steps described here are based on the following example topology (based on the Cumulus Networks Reference Topology):

This topology:

The BGP EVPN configuration for a centralized routing topology is slightly different on the exit/routing leafs compared to the other ToR leaf switches.

  1. Run the following NCLU commands on each type of device shown (leaf, exit, spine):

    Leaf node NCLU commands

# BGP changes
cumulus@switch:~$ net add bgp l2vpn evpn neighbor swp51-52 activate
cumulus@switch:~$ net add bgp l2vpn evpn advertise-all-vni

# Disable MAC learning on VNI
cumulus@switch:~$ net add vxlan vni-13 bridge learning off
cumulus@switch:~$ net add vxlan vni-24 bridge learning off

# Remove LNV (vxrd) configuration
cumulus@switch:~$ net del loopback lo vxrd-src-ip
cumulus@switch:~$ net del loopback lo vxrd-svcnode-ip
**Exit node NCLU commands**
# BGP changes
cumulus@switch:~$ net add bgp l2vpn evpn neighbor swp51-52 activate
cumulus@switch:~$ net add bgp l2vpn evpn advertise-all-vni
cumulus@switch:~$ net add bgp l2vpn evpn advertise-default-gw

# Disable MAC learning on VNI
cumulus@switch:~$ net add vxlan vni-13 bridge learning off
cumulus@switch:~$ net add vxlan vni-24 bridge learning off

# Remove LNV (vxrd) configuration
cumulus@switch:~$ net del loopback lo vxrd-src-ip
cumulus@switch:~$ net del loopback lo vxrd-svcnode-ip
**Spine node NCLU commands**
# BGP changes
cumulus@switch:~$ net add bgp l2vpn evpn neighbor swp1-4 activate

Remove LNV service node (vxsnd) configuration
cumulus@switch:~$ net del lnv service-node anycast-ip 10.0.0.200
cumulus@switch:~$ net del lnv service-node peers 10.0.0.21 10.0.0.22
cumulus@switch:~$ net del lnv service-node source [primary-loopback-ip]

# Remove unused LNV anycast address 10.0.0.200
cumulus@switch:~$ net del loopback lo ip address 10.0.0.200/32
cumulus@switch:~$ net del bgp ipv4 unicast network 10.0.0.200/32
  1. Manually disable and stop the LNV daemons. NCLU can remove the LNV configuration from the configuration files, but you must manually stop and disable these daemons before you commit the NCLU changes. After you commit the NCLU changes, NCLU restarts the BGP daemon, which enables the EVPN address family.

    Traffic loss can start to occur at this point.

  2. To disable and stop the LNV registration daemon, run the following commands on the leaf and exit nodes:

cumulus@switch:~$ sudo systemctl disable vxrd
cumulus@switch:~$ sudo systemctl stop vxrd
  1. To disable and stop the LNV service node daemon, run the following commands on the spine nodes:
cumulus@switch:~$ sudo systemctl disable vxsnd
cumulus@switch:~$ sudo systemctl stop vxsnd
  1. To commit and apply the pending NCLU changes, run the following command on all the nodes:
cumulus@switch:~$ net commit

Verify the Upgrade

To check that LNV is disabled, run the net show lnv command on any node. This command returns no output when LNV is disabled.

This command is for verification on Cumulus Linux 3.x only. This command has been removed in Cumulus Linux 4.0 and does not work after you upgrade.

cumulus@switch:~$ net show lnv

To ensure that EVPN BGP neighbors are up, run the net show bgp l2vpn summary command:

cumulus@switch:~$ net show bgp l2vpn evpn summary
BGP router identifier 10.0.0.11, local AS number 65011 vrf-id 0
BGP table version 0
RIB entries 23, using 3496 bytes of memory
Peers 2, using 39 KiB of memory
Neighbor        V         AS MsgRcvd MsgSent   TblVer  InQ OutQ  Up/Down State/PfxRcd
spine01(swp51)  4      65020   10932   11064        0    0    0 00:14:28           48
spine02(swp52)  4      65020   10938   11068        0    0    0 00:14:27           48
Total number of neighbors 2

To examine the EVPN routes, run the net show bgp l2vpn evpn route command. Because a MAC address only appears as a type-2 route if the host has generated traffic and its MAC is learned by the local EVPN-enabled switch, a host that does not send any traffic does not create a type-2 EVPN route until it sends a frame that ingresses the EVPN-enabled local switch.

cumulus@switch:~$ net show bgp l2vpn evpn route  
BGP table version is 45, local router ID is 10.0.0.11
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal
Origin codes: i - IGP, e - EGP, ? - incomplete
EVPN type-2 prefix: [2]:[ESI]:[EthTag]:[MAClen]:[MAC]:[IPlen]:[IP]
EVPN type-3 prefix: [3]:[EthTag]:[IPlen]:[OrigIP]
EVPN type-5 prefix: [5]:[ESI]:[EthTag]:[IPlen]:[IP]
    Network          Next Hop            Metric LocPrf Weight Path
Route Distinguisher: 10.0.0.11:2
*> [2]:[0]:[0]:[48]:[00:03:00:11:11:01]
                    10.0.0.100                         32768 i
*> [2]:[0]:[0]:[48]:[02:03:00:11:11:01]
                    10.0.0.100                         32768 i
*> [2]:[0]:[0]:[48]:[02:03:00:11:11:02]
                    10.0.0.100                         32768 i
*> [3]:[0]:[32]:[10.0.0.100]
                    10.0.0.100                         32768 i
Route Distinguisher: 10.0.0.11:3
*> [2]:[0]:[0]:[48]:[00:03:00:22:22:02]
                    10.0.0.100                         32768 i
*> [2]:[0]:[0]:[48]:[02:03:00:22:22:01]
                    10.0.0.100                         32768 i
*> [2]:[0]:[0]:[48]:[02:03:00:22:22:02]
                    10.0.0.100                         32768 i
*> [3]:[0]:[32]:[10.0.0.100]
                    10.0.0.100                         32768 i
Route Distinguisher: 10.0.0.13:2
*  [2]:[0]:[0]:[48]:[00:03:00:33:33:01]
                    10.0.0.101                             0 65020 65013 i
*> [2]:[0]:[0]:[48]:[00:03:00:33:33:01]
                    10.0.0.101                             0 65020 65013 i
*  [2]:[0]:[0]:[48]:[02:03:00:33:33:01]
                    10.0.0.101                             0 65020 65013 i
*> [2]:[0]:[0]:[48]:[02:03:00:33:33:01]
                    10.0.0.101                             0 65020 65013 i
*  [2]:[0]:[0]:[48]:[02:03:00:33:33:02]
                    10.0.0.101                             0 65020 65013 i
*> [2]:[0]:[0]:[48]:[02:03:00:33:33:02]
                    10.0.0.101                             0 65020 65013 i
*  [3]:[0]:[32]:[10.0.0.101]
                    10.0.0.101                             0 65020 65013 i
*> [3]:[0]:[32]:[10.0.0.101]
                    10.0.0.101                             0 65020 65013 i
...

You can filter the EVPN route output by route type. The multicast route type corresponds to type-3. The prefix route type is type-5 (but is not used here).

cumulus@switch:~$ net show bgp l2vpn evpn route type 
   macip      : MAC-IP (Type-2) route
   multicast  : Multicast
   prefix     : An IPv4 or IPv6 prefix

In the EVPN route output below, Cumulus Linux learned 00:03:00:33:33:01 with a next-hop (VTEP IP address) of 10.0.0.101. The MAC address of server03 is 00:03:00:33:33:01.

cumulus@leaf01:~$ net show bgp l2vpn evpn route
...

Route Distinguisher: 10.0.0.13:2
*  [2]:[0]:[0]:[48]:[00:03:00:33:33:01]
                     10.0.0.101                             0 65020 65013 i
...

To ensure the type-2 route is installed in the bridge table, run the net show bridge macs <mac-address> command on leaf01:

cumulus@leaf01:~$ net show bridge macs 00:03:00:33:33:01
VLAN      Master  Interface  MAC                TunnelDest  State  Flags          LastSeen
--------  ------  ---------  -----------------  ----------  -----  -------------  --------
13        bridge  vni-13     00:03:00:33:33:01                     offload        00:01:49
untagged          vni-13     00:03:00:33:33:01  10.0.0.101         self, offload  00:01:49

Back up and Restore

You can back up the current configuration on a switch and restore the configuration on the same switch or on another Cumulus Linux switch of the same type and release. The backup is a compressed tar file that includes all configuration files installed by Debian packages and marked as configuration files. In addition, the backup contains files in the /etc directory that are not installed by a Debian package but are modified when you install a new image or enable/disable certain services (such as the Cumulus license file).

Cumulus Linux automatically creates a backup of the configuration files on the switch after you install the Cumulus Linux image, in case you want to return to the initial switch configuration. NCLU automatically creates a backup of the configuration files when you run the net commit command and restores a previous configuration when you run the net rollback command.

Back up Configuration Files

To back up the current configuration files on the switch, run the config-backup command:

cumulus@switch:~$ sudo config-backup

If you run this command without any options, Cumulus Linux creates a backup of the current configuration and stores the backup file in the /var/lib/config-backup/backups directory. The filename includes the date and time you run the backup, and the switch name; for example, config_backup-2019-04-23-21.30.47_leaf01. You can restore the backup with the config-restore command, described below.

The switch can store up to 30 non-permanent backup files (or can allocate a maximum of 25 MB of disc space) in addition to the permanent backup files (see the -p option below). When this limit is reached, Cumulus Linux keeps the oldest and the newest backup files, then starts removing the second oldest file up to the second newest file.

Cumulus Linux recommends you copy the backup file off the switch after backup is complete.

The config-backup command includes the following options:

OptionDescription
-hDisplays this list of command options.
-dEnables debugging output, which shows status messages during the backup process.
-D <description>Adds a description, which is shown in the archive file list when you run the config-restore -l command.
-pAdds -perm to the end of the backup filename to mark it as permanent. For example, config_backup-2019-04-23-21.30.47_leaf01-perm. Be careful when using this option. Permanent backup files are not removed.
-qRuns the command in quiet mode. No status messages are shown, only errors.
-t <type>Specifies the type of configuration, which is shown in the archive file list when you run the config-restore -l command. You can provide any short text. For example, you can specify pre, post, or pre-restore.
-vEnables verbose mode to show messages during the backup process.
-X <pattern>Excludes certain files that match a specified pattern. For example, to exclude all backup files ending with a tilde (~), use the -X .*~$ option.

config-backup Command Examples

The following command example creates a backup file in debugging mode and provides the description myconfig, which shows in the backup archive list.

cumulus@switch:~$ sudo config-backup -d -D myconfig 

The following command example creates a backup file in quiet mode and excludes files that end in a tilde (~).

cumulus@switch:~$ sudo config-backup -q -X .*~$

The following command example creates a backup file in verbose mode and marks the file as permanent.

cumulus@switch:~$ sudo config-backup -pv

Restore Backup Files

You can restore a backup to the same switch or to a different switch. When restoring to a different switch, the switch must be of the same type and release. For example, you can restore a backup from a Broadcom Trident3 switch to a Broadcom Trident3 switch; however, you cannot restore a backup from a Broadcom Trident3 switch to a Mellanox Spectrum or to a Broadcom Tomahawk2 switch.

To restore a backup file, run the config-restore command with a specific filename (-b <filename>), file number (-n <number>), or the -N option, which restores the most recent backup file.

You can run the config-restore -l command to list the archived backup files by filename and number (see config-restore Command Examples below).

cumulus@switch:~$ sudo config-restore -b config_backup-2019-04-23-21.30.47_leaf01
cumulus@switch:~$ sudo config-restore -n 10
cumulus@switch:~$ sudo config-restore -N

After the backup file is restored successfully, you are prompted to restart any affected services or reboot the switch if necessary.

Cumulus Linux reports any issues encountered during restore and prompts you to continue or stop.

  • The config-restore command requires a filename, file number, or the most recent file option (-N).
  • You can only run one config-backup or config-restore command instance at the same time.

The config-restore command includes the following options:

OptionDescription
-hDisplays this list of command options.
-a <directory>Restores the backup to the directory specified.
-BRuns no backup before restoring the configuration. If you do not specify this option, Cumulus Linux runs a backup to save the current configuration before the restore so that you can do a rollback if needed.
-b <filename>Specifies the name of the backup file you want to restore (shown by -l).
-DShows the differences between the current configuration and the configuration in the backup file.
-dDisplays debugging output, which provides status messages during the restore process.
-fForces the restore; does not prompt for confirmations.
-F <filename>Shows differences for only this file (used with -D).
-iDisplays information about the current backup file.
-LLists the configuration files in the backup file.
-lLists all backup files archived on the switch and includes the file number, type, and description.
-NRestores the newest (most recent) backup file.
-n <number>Specifies the backup file by number (shown by -l).
-qRuns the command in quiet mode. No status messages are displayed, only errors.
-TRuns the command in test mode; does not restore the configuration but shows what would be restored.
-vEnables verbose mode to display status messages during restore.

config-restore Command Examples

The following command example lists the backup files available on the switch. The list includes the file number (#), type, description, and filename. Type is the text specified with the config-backup -t option.

cumulus@switch:~$ sudo config-restore -l
# Type       Description               Name
1 Initial    First system boot         config_backup-2019-04-23-00.42.11_cumulus-perm
2 Initial    First system boot         config_backup-2019-04-23-00.47.43_cumulus-perm
3 Initial    First system boot         config_backup-2019-04-23-18.12.26_cumulus-perm
4 pre nclu "net commit" (user cumulus) config_backup-2019-04-23-19.55.13_leaf01
5 post-4     nclu "net commit" (user cumulus)   config_backup-2019-04-23-19.55.26_leaf01
6            config_backup-2019-04-23-21.20.41_leaf01
7            config_backup-2019-04-23-21.30.47_leaf01-perm
...

The following command example runs in verbose mode to restore the backup file config_backup-2019-04-23-21.30.47_leaf01.

cumulus@switch:~$ sudo config-restore -v -b config_backup-2019-04-23-21.30.47_leaf01

The following command example runs test mode to restore the most recent backup file (no configuration is actually restored).

cumulus@switch:~$ sudo config-restore -T -N

The following command example lists the files in the most recent backup file.

cumulus@switch:~$ sudo config-restore -L -N

Adding and Updating Packages

You use the Advanced Packaging Tool (apt) to manage additional applications (in the form of packages) and to install the latest updates.

Updating, upgrading, and installing packages with apt causes disruptions to network services:

  • Upgrading a package might result in services being restarted or stopped as part of the upgrade process.
  • Installing a package might disrupt core services by changing core service dependency packages. In some cases, installing new packages might also upgrade additional existing packages due to dependencies.

If services are stopped, you might need to reboot the switch for those services to restart.

Update the Package Cache

To work properly, apt relies on a local cache listing of the available packages. You must populate the cache initially, then periodically update it with sudo -E apt-get update:

cumulus@switch:~$ sudo -E apt-get update
Get:1 http://apt.cumulusnetworks.com CumulusLinux-4-latest InRelease [7,624 B]
Get:2 http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest InRelease [7,555 B]
Get:3 http://apt.cumulusnetworks.com CumulusLinux-4-latest-updates InRelease [7,660 B]
Get:4 http://apt.cumulusnetworks.com CumulusLinux-4-latest/cumulus Sources [20 B]
Get:5 http://apt.cumulusnetworks.com CumulusLinux-4-latest/upstream Sources [20 B]
Get:6 http://apt.cumulusnetworks.com CumulusLinux-4-latest/cumulus amd64 Packages [38.4 kB]
Get:7 http://apt.cumulusnetworks.com CumulusLinux-4--latest/upstream amd64 Packages [445 kB]
Get:8 http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/cumulus Sources [20 B]
Get:9 http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/upstream Sources [11.8 kB]
Get:10 http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/cumulus amd64 Packages [20 B]
Get:11 http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/upstream amd64 Packages [8,941 B]
Get:12 http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/cumulus Sources [20 B]
Get:13 http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/upstream Sources [776 B]
Get:14 http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/cumulus amd64 Packages [38.4 kB]
Get:15 http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/upstream amd64 Packages [444 kB]
Ign http://apt.cumulusnetworks.com CumulusLinux-4-latest/cumulus Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-latest/cumulus Translation-en
Ign http://apt.cumulusnetworks.com CumulusLinux-4-latest/upstream Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-latest/upstream Translation-en
Ign http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/cumulus Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/cumulus Translation-en
Ign http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/upstream Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-security-updates-latest/upstream Translation-en
Ign http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/cumulus Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/cumulus Translation-en
Ign http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/upstream Translation-en_US
Ign http://apt.cumulusnetworks.com CumulusLinux-4-updates-latest/upstream Translation-en
Fetched 1,011 kB in 1s (797 kB/s)
Reading package lists... Done

Cumulus Networks recommends you use the -E option with sudo whenever you run any apt-get command. This option preserves your environment variables (such as HTTP proxies) before you install new packages or upgrade your distribution.

List Available Packages

After the cache is populated, use the apt-cache command to search the cache and find the packages in which you are interested or to get information about an available package.

Here are examples of the search and show sub-commands:

cumulus@switch:~$ apt-cache search tcp
collectd-core - statistics collection and monitoring daemon (core system)
fakeroot - tool for simulating superuser privileges
iperf - Internet Protocol bandwidth measuring tool
iptraf-ng - Next Generation Interactive Colorful IP LAN Monitor
libfakeroot - tool for simulating superuser privileges - shared libraries
libfstrm0 - Frame Streams (fstrm) library
libibverbs1 - Library for direct userspace use of RDMA (InfiniBand/iWARP)
libnginx-mod-stream - Stream module for Nginx
libqt4-network - Qt 4 network module
librtr-dev - Small extensible RPKI-RTR-Client C library - development files
librtr0 - Small extensible RPKI-RTR-Client C library
libwiretap8 - network packet capture library -- shared library
libwrap0 - Wietse Venema's TCP wrappers library
libwrap0-dev - Wietse Venema's TCP wrappers library, development files
netbase - Basic TCP/IP networking system
nmap-common - Architecture independent files for nmap
nuttcp - network performance measurement tool
openssh-client - secure shell (SSH) client, for secure access to remote machines
openssh-server - secure shell (SSH) server, for secure access from remote machines
openssh-sftp-server - secure shell (SSH) sftp server module, for SFTP access from remote machines
python-dpkt - Python 2 packet creation / parsing module for basic TCP/IP protocols
rsyslog - reliable system and kernel logging daemon
socat - multipurpose relay for bidirectional data transfer
tcpdump - command-line network traffic analyzer
cumulus@switch:~$ apt-cache show tcpdump
Package: tcpdump
Version: 4.9.3-1~deb10u1
Installed-Size: 1109
Maintainer: Romain Francoise <rfrancoise@debian.org>
Architecture: amd64
Replaces: apparmor-profiles-extra (<< 1.12~)
Depends: libc6 (>= 2.14), libpcap0.8 (>= 1.5.1), libssl1.1 (>= 1.1.0)
Suggests: apparmor (>= 2.3)
Breaks: apparmor-profiles-extra (<< 1.12~)
Size: 400060
SHA256: 3a63be16f96004bdf8848056f2621fbd863fadc0baf44bdcbc5d75dd98331fd3
SHA1: 2ab9f0d2673f49da466f5164ecec8836350aed42
MD5sum: 603baaf914de63f62a9f8055709257f3
Description: command-line network traffic analyzer
 This program allows you to dump the traffic on a network. tcpdump
 is able to examine IPv4, ICMPv4, IPv6, ICMPv6, UDP, TCP, SNMP, AFS
 BGP, RIP, PIM, DVMRP, IGMP, SMB, OSPF, NFS and many other packet
 types.
 .
 It can be used to print out the headers of packets on a network
 interface, filter packets that match a certain expression. You can
 use this tool to track down network problems, to detect attacks
 or to monitor network activities.
Description-md5: f01841bfda357d116d7ff7b7a47e8782
Homepage: http://www.tcpdump.org/
Multi-Arch: foreign
Section: net
Priority: optional
Filename: pool/upstream/t/tcpdump/tcpdump_4.9.3-1~deb10u1_amd64.deb

The search commands look for the search terms not only in the package name but in other parts of the package information; the search matches on more packages than you might expect.

List Packages Installed on the System

The the apt-cache command shows information about all the packages available in the repository. To see which packages are actually installed on your system with their versions, run the following commands.

NCLU Commands

Run the net show package version command:

cumulus@switch:~$ net show package version
Package                            Installed Version(s)
---------------------------------  -----------------------------------------------------------------------
acpi                               1.7-1.1
acpi-support-base                  0.142-8
acpid                              1:2.0.31-1
adduser                            3.118
apt                                1.8.2
arping                             2.19-6
arptables                          0.0.4+snapshot20181021-4
atftp                              0.7.git20120829-3.1
atftpd                             0.7.git20120829-3.1
...
Linux Commands

Run the dpkg -l command:

cumulus@switch:~$ dpkg -l
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                Version                   Architecture Description
+++-===================-=========================-============-=================================
ii  acpi                1.7-1.1                   amd64        displays information on ACPI devices
ii  acpi-support-base   0.142-8                   all          scripts for handling base ACPI events such as th
ii  acpid               1:2.0.31-1                amd64        Advanced Configuration and Power Interface event
ii  adduser             3.118                     all          add and remove users and groups
ii  apt                 1.8.2                     amd64        commandline package manager
ii  arping              2.19-6                    amd64        sends IP and/or ARP pings (to the MAC address)
ii  arptables           0.0.4+snapshot20181021-4  amd64        ARP table administration
ii  atftp               0.7.git20120829-3.1       amd64        advanced TFTP client
ii  atftpd              0.7.git20120829-3.1       amd64        advanced TFTP server
...

The apps repository was removed in Cumulus Linux 4.0.0.

Show the Version of a Package

To show the version of a specific package installed on the system:

NCLU Commands

Run the net show package version <package> command. For example, the following command shows which version of the vrf package is installed on the system:

cumulus@switch:~$ net show package version vrf
1.0-cl4u1
Linux Commands

Run the Linux dpkg -l <package_name> command. For example, the following command shows which version of the vrf package is installed on the system:

cumulus@switch:~$ dpkg -l vrf
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name       Version      Architecture Description
+++-==========-============-============-=================================
ii  vrf        1.0-cl4u1    amd64        Linux tools for VRF

Upgrade Packages

You cannot upgrade to Cumulus Linux 4.0 from Cumulus 3.7 by upgrading packages. You must install a disk image using ONIE. Refer to Upgrading Cumulus Linux.

To upgrade all the packages installed on the system to their latest versions, run the following commands:

cumulus@switch:~$ sudo -E apt-get update
cumulus@switch:~$ sudo -E apt-get upgrade

A list of packages that will be upgraded is displayed and you are prompted to continue.

The above commands upgrade all installed versions with their latest versions but do not install any new packages.

Add Packages from Another Repository

As shipped, Cumulus Linux searches the Cumulus Linux repository for available packages. You can add additional repositories to search by adding them to the list of sources that apt-get consults. See man sources.list for more information.

Cumulus Networks has added features or made bug fixes to certain packages; you must not replace these packages with versions from other repositories. Cumulus Linux is configured to ensure that the packages from the Cumulus Linux repository are always preferred over packages from other repositories.

If you want to install packages that are not in the Cumulus Linux repository, the procedure is the same as above, but with one additional step.

Packages that are not part of the Cumulus Linux Repository are not typically tested and might not be supported by Cumulus Linux Technical Support.

Installing packages outside of the Cumulus Linux repository requires the use of sudo -E apt-get; however, depending on the package, you can use easy-install and other commands.

To install a new package, complete the following steps:

  1. Run the dpkg command to ensure that the package is not already installed on the system:

    cumulus@switch:~$ dpkg -l | grep {name of package}
    
  2. If the package is installed already, ensure it is the version you need. If it is an older version, update the package from the Cumulus Linux repository:

    cumulus@switch:~$ sudo -E apt-get update
    cumulus@switch:~$ sudo -E apt-get install {name of package}
    cumulus@switch:~$ sudo -E apt-get upgrade
    
  3. If the package is not on the system, the package source location is most likely not in the /etc/apt/sources.list file. If the source for the new package is not in sources.list, edit and add the appropriate source to the file. For example, add the following if you want a package from the Debian repository that is not in the Cumulus Linux repository:

    deb http://http.us.debian.org/debian buster main
    deb http://security.debian.org/ buster/updates main
    

    Otherwise, the repository might be listed in /etc/apt/sources.list but is commented out, as can be the case with the early-access repository:

    #deb http://apt.cumulusnetworks.com/repo CumulusLinux-4-early-access cumulus
    

    To uncomment the repository, remove the # at the start of the line, then save the file:

    deb http://apt.cumulusnetworks.com/repo CumulusLinux-4-early-access cumulus
    
  4. Run sudo -E apt-get update, then install the package and upgrade:

    cumulus@switch:~$ sudo -E apt-get update
    cumulus@switch:~$ sudo -E apt-get install {name of package}
    cumulus@switch:~$ sudo -E apt-get upgrade
    

Add Packages from the Cumulus Linux Local Archive

Cumulus Linux contains a local archive embedded in the Cumulus Linux disk image. This archive contains the packages needed to install ifplugd, LDAP, RADIUS or TACACS+ without needing a network connection.

The archive is called cumulus-local-apt-archive and is referenced in the /etc/apt/cumulus-local-apt-archive-sources.list file. It contains the following packages:

You add these packages normally with apt-get update && apt-get install, as described above.

Caveats and Errata

At this time, you cannot directly browse the contents of the apt.cumulusnetworks.com repository using HTTP.

Zero Touch Provisioning - ZTP

Zero touch provisioning (ZTP) enables you to deploy network devices quickly in large-scale environments. On first boot, Cumulus Linux invokes ZTP, which executes the provisioning automation used to deploy the device for its intended role in the network.

The provisioning framework allows for a one-time, user-provided script to be executed. You can develop this script using a variety of automation tools and scripting languages, providing ample flexibility for you to design the provisioning scheme to meet your needs. You can also use it to add the switch to a configuration management (CM) platform such as Puppet, Chef, CFEngine or possibly a custom, proprietary tool.

While developing and testing the provisioning logic, you can use the ztp command in Cumulus Linux to manually invoke your provisioning script on a device.

ZTP in Cumulus Linux can occur automatically in one of the following ways, in this order:

Each method is discussed in greater detail below.

Use a Local File

ZTP only looks once for a ZTP script on the local file system when the switch boots. ZTP searches for an install script that matches an ONIE-style waterfall in /var/lib/cumulus/ztp, looking for the most specific name first, and ending at the most generic:

For example:

cumulus-ztp-amd64-cel_pebble-rUNKNOWN
cumulus-ztp-amd64-cel_pebble
cumulus-ztp-cel_pebble
cumulus-ztp-amd64
cumulus-ztp

You can also trigger the ZTP process manually by running the ztp --run <URL> command, where the URL is the path to the ZTP script.

Use a USB Drive

This feature has been tested only with thumb drives, not an actual external large USB hard drive.

If the ztp process does not discover a local script, it tries once to locate an inserted but unmounted USB drive. If it discovers one, it begins the ZTP process.

Cumulus Linux supports the use of a FAT32, FAT16, or VFAT-formatted USB drive as an installation source for ZTP scripts. You must plug in the USB drive before you power up the switch.

At minimum, the script must:

Follow these steps to perform ZTP using a USB drive:

  1. Copy the Cumulus Linux license and installation image to the USB drive.
  2. The ztp process searches the root filesystem of the newly mounted drive for filenames matching an ONIE-style waterfall (see the patterns and examples above), looking for the most specific name first, and ending at the most generic.
  3. The contents of the script are parsed to ensure it contains the CUMULUS-AUTOPROVISIONING flag (see example scripts).

The USB drive is mounted to a temporary directory under /tmp (for example, /tmp/tmpigGgjf/). To reference files on the USB drive, use the environment variable ZTP_USB_MOUNTPOINT to refer to the USB root partition.

ZTP over DHCP

If the ztp process does not discover a local/ONIE script or applicable USB drive, it checks DHCP every ten seconds for up to five minutes for the presence of a ZTP URL specified in /var/run/ztp.dhcp. The URL can be any of HTTP, HTTPS, FTP or TFTP.

For ZTP using DHCP, provisioning initially takes place over the management network and is initiated through a DHCP hook. A DHCP option is used to specify a configuration script. This script is then requested from the Web server and executed locally on the switch.

The ZTP process over DHCP follows these steps:

  1. The first time you boot Cumulus Linux, eth0 is configured for DHCP and makes a DHCP request.
  2. The DHCP server offers a lease to the switch.
  3. If option 239 is present in the response, the ZTP process starts.
  4. The ZTP process requests the contents of the script from the URL, sending additional HTTP headers containing details about the switch.
  5. The contents of the script are parsed to ensure it contains the CUMULUS-AUTOPROVISIONING flag (see example scripts).
  6. If provisioning is necessary, the script executes locally on the switch with root privileges.
  7. The return code of the script is examined. If it is 0, the provisioning state is marked as complete in the autoprovisioning configuration file.

Trigger ZTP over DHCP

If provisioning has not already occurred, it is possible to trigger the ZTP process over DHCP when eth0 is set to use DHCP and one of the following events occur:

You can also run the ztp --run <URL> command, where the URL is the path to the ZTP script.

Configure the DHCP Server

During the DHCP process over eth0, Cumulus Linux requests DHCP option 239. This option is used to specify the custom provisioning script.

For example, the /etc/dhcp/dhcpd.conf file for an ISC DHCP server looks like:

option cumulus-provision-url code 239 = text;

  subnet 192.0.2.0 netmask 255.255.255.0 {
  range 192.0.2.100 192.168.0.200;
  option cumulus-provision-url "http://192.0.2.1/demo.sh";
}

Additionally, you can specify the hostname of the switch with the host-name option:

subnet 192.168.0.0 netmask 255.255.255.0 {
  range 192.168.0.100 192.168.0.200;
  option cumulus-provision-url "http://192.0.2.1/demo.sh";
  host dc1-tor-sw1 { hardware ethernet 44:38:39:00:1a:6b; fixed-address 192.168.0.101; option host-name "dc1-tor-sw1"; }
}

Do not use an underscore (_) in the hostname; underscores are not permitted in hostnames.

Inspect HTTP Headers

The following HTTP headers are sent in the request to the webserver to retrieve the provisioning script:

Header                        Value                 Example
------                        -----                 -------
User-Agent                                          CumulusLinux-AutoProvision/0.4
CUMULUS-ARCH                  CPU architecture      x86_64
CUMULUS-BUILD                                       4.0.0-5c6829a-201309251712-final
CUMULUS-LICENSE-INSTALLED     Either 0 or 1         1
CUMULUS-MANUFACTURER                                odm
CUMULUS-PRODUCTNAME                                 switch_model
CUMULUS-SERIAL                                      XYZ123004
CUMULUS-BASE-MAC                                    44:38:39:FF:40:94
CUMULUS-MGMT-MAC                                    44:38:39:FF:00:00
CUMULUS-VERSION                                     4.0.0
CUMULUS-PROV-COUNT                                  0
CUMULUS-PROV-MAX                                    32

Write ZTP Scripts

Remember to include the following line in any of the supported scripts that you expect to run using the autoprovisioning framework.

# CUMULUS-AUTOPROVISIONING

This line is required somewhere in the script file for execution to occur.

The script must contain the CUMULUS-AUTOPROVISIONING flag. You can include this flag in a comment or remark; the flag does not need to be echoed or written to stdout.

You can write the script in any language currently supported by Cumulus Linux, such as:

The script must return an exit code of 0 upon success, as this triggers the autoprovisioning process to be marked as complete in the autoprovisioning configuration file.

The following script installs Cumulus Linux and its license from a USB drive and applies a configuration:

#!/bin/bash
function error() {
  echo -e "\e[0;33mERROR: The ZTP script failed while running the command $BASH_COMMAND at line $BASH_LINENO.\e[0m" >&2
  exit 1
}

# Log all output from this script
exec >> /var/log/autoprovision 2>&1
date "+%FT%T ztp starting script $0"

trap error ERR

#Add Debian Repositories
echo "deb http://http.us.debian.org/debian buster main" >> /etc/apt/sources.list
echo "deb http://security.debian.org/ buster/updates main" >> /etc/apt/sources.list

#Update Package Cache
apt-get update -y

#Load interface config from usb
cp ${ZTP_USB_MOUNTPOINT}/interfaces /etc/network/interfaces

#Load port config from usb
#   (if breakout cables are used for certain interfaces)
cp ${ZTP_USB_MOUNTPOINT}/ports.conf /etc/cumulus/ports.conf

#Install a License from usb and restart switchd
/usr/cumulus/bin/cl-license -i ${ZTP_USB_MOUNTPOINT}/license.txt && systemctl restart switchd.service

#Reload interfaces to apply loaded config
ifreload -a

#Output state of interfaces
net show interface

# CUMULUS-AUTOPROVISIONING
exit 0

Several ZTP example scripts are available in the Cumulus GitHub repository.

Best Practices

ZTP scripts come in different forms and frequently perform many of the same tasks. As BASH is the most common language used for ZTP scripts, the following BASH snippets are provided to accelerate your ability to perform common tasks with robust error checking.

Install a License

Use the following function to include error checking for license file installation.

function install_license(){
     # Install license
     echo "$(date) INFO: Installing License..."
     echo $1 | /usr/cumulus/bin/cl-license -i
     return_code=$?
     if [ "$return_code" == "0" ]; then
         echo "$(date) INFO: License Installed."
     else
         echo "$(date) ERROR: License not installed. Return code was: $return_code"
         /usr/cumulus/bin/cl-license
         exit 1
     fi
}

Test DNS Name Resolution

DNS names are frequently used in ZTP scripts. The ping_until_reachable function tests that each DNS name resolves into a reachable IP address. Call this function with each DNS target used in your script before you use the DNS name elsewhere in your script.

The following example shows how to call the ping_until_reachable function in the context of a larger task.

function ping_until_reachable(){
    last_code=1
    max_tries=30
    tries=0
    while [ "0" != "$last_code" ] && [ "$tries" -lt "$max_tries" ]; do
        tries=$((tries+1))
        echo "$(date) INFO: ( Attempt $tries of $max_tries ) Pinging $1 Target Until Reachable."
        ping $1 -c2 &> /dev/null
        last_code=$?
            sleep 1
    done
    if [ "$tries" -eq "$max_tries" ] && [ "$last_code" -ne "0" ]; then
        echo "$(date) ERROR: Reached maximum number of attempts to ping the target $1 ."
        exit 1
    fi
}

Check the Cumulus Linux Release

The following script segment demonstrates how to check which Cumulus Linux release is running currently and upgrades the node if the release is not the target release. If the release is the target release, normal ZTP tasks execute. This script calls the ping_until_reachable script (described above) to make sure the server holding the image server and the ZTP script is reachable.

function init_ztp(){
    #do normal ZTP tasks
}

CUMULUS_TARGET_RELEASE=3.5.3
CUMULUS_CURRENT_RELEASE=$(cat /etc/lsb-release  | grep RELEASE | cut -d "=" -f2)
IMAGE_SERVER_HOSTNAME=webserver.example.com
IMAGE_SERVER= "http:// "$IMAGE_SERVER_HOSTNAME "/ "$CUMULUS_TARGET_RELEASE ".bin "
ZTP_URL= "http:// "$IMAGE_SERVER_HOSTNAME "/ztp.sh "

if [ "$CUMULUS_TARGET_RELEASE" != "$CUMULUS_CURRENT_RELEASE" ]; then
    ping_until_reachable $IMAGE_SERVER_HOSTNAME
    /usr/cumulus/bin/onie-install -fa -i $IMAGE_SERVER -z $ZTP_URL && reboot
else
    init_ztp && reboot
fi
exit 0

Apply Management VRF Configuration

If you apply a management VRF in your script, either apply it last or reboot instead. If you do not apply a management VRF last, you need to prepend any commands that require eth0 to communicate out with /usr/bin/ip vrf exec mgmt; for example, /usr/bin/ip vrf exec mgmt apt-get update -y.

Perform Ansible Provisioning Callbacks

After initially configuring a node with ZTP, use Provisioning Callbacks to inform Ansible Tower or AWX that the node is ready for more detailed provisioning. The following example demonstrates how to use a provisioning callback:

/usr/bin/curl -H "Content-Type:application/json" -k -X POST --data '{"host_config_key":"'somekey'"}' -u username:password http://ansible.example.com/api/v2/job_templates/1111/callback/

Disable the DHCP Hostname Override Setting

Make sure to disable the DHCP hostname override setting in your script (NCLU does this automatically).

function set_hostname(){
    # Remove DHCP Setting of Hostname
    sed s/'SETHOSTNAME="yes"'/'SETHOSTNAME="no"'/g -i /etc/dhcp/dhclient-exit-hooks.d/dhcp-sethostname
    hostnamectl set-hostname $1
}

NCLU in ZTP Scripts

Not all aspects of NCLU are supported when running during ZTP. Use traditional Linux methods of providing configuration to the switch during ZTP.

When you use NCLU in ZTP scripts, add the following loop to make sure NCLU has time to start up before being called.

# Waiting for NCLU to finish starting up
last_code=1
while [ "1" == "$last_code" ]; do
    net show interface &> /dev/null
    last_code=$?
done

net add vrf mgmt
net add time zone Etc/UTC
net add time ntp server 192.168.0.254 iburst
net commit

Test ZTP Scripts

There are a few commands you can use to test and debug your ZTP scripts.

You can use verbose mode to debug your script and see where your script failed. Include the -v option when you run ZTP:

cumulus@switch:~$ sudo ztp -v -r http://192.0.2.1/demo.sh
Attempting to provision via ZTP Manual from http://192.0.2.1/demo.sh

Broadcast message from root@dell-s6010-01 (ttyS0) (Tue May 10 22:44:17 2016):  

ZTP: Attempting to provision via ZTP Manual from http://192.0.2.1/demo.sh
ZTP Manual: URL response code 200
ZTP Manual: Found Marker CUMULUS-AUTOPROVISIONING
ZTP Manual: Executing http://192.0.2.1/demo.sh
error: ZTP Manual: Payload returned code 1
error: Script returned failure

To see if ZTP is enabled and to see results of the most recent execution, you can run the ztp -s command.

cumulus@switch:~$ ztp -s
ZTP INFO:

State              enabled
Version            1.0
Result             Script Failure
Date               Mon 20 May 2019 09:31:27 PM UTC
Method             ZTP DHCP
URL                http://192.0.2.1/demo.sh

If ZTP runs when the switch boots and not manually, you can run the systemctl -l status ztp.service then journalctl -l -u ztp.service to see if any failures occur:

cumulus@switch:~$ sudo systemctl -l status ztp.service
● ztp.service - Cumulus Linux ZTP
    Loaded: loaded (/lib/systemd/system/ztp.service; enabled)
    Active: failed (Result: exit-code) since Wed 2016-05-11 16:38:45 UTC; 1min 47s ago
        Docs: man:ztp(8)
    Process: 400 ExecStart=/usr/sbin/ztp -b (code=exited, status=1/FAILURE)
    Main PID: 400 (code=exited, status=1/FAILURE)

May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP USB: Device not found
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Looking for ZTP Script provided by DHCP
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: Attempting to provision via ZTP DHCP from http://192.0.2.1/demo.sh
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: URL response code 200
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Found Marker CUMULUS-AUTOPROVISIONING
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Executing http://192.0.2.1/demo.sh
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Payload returned code 1
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: Script returned failure
May 11 16:38:45 dell-s6010-01 systemd[1]: ztp.service: main process exited, code=exited, status=1/FAILURE
May 11 16:38:45 dell-s6010-01 systemd[1]: Unit ztp.service entered failed state.
cumulus@switch:~$
cumulus@switch:~$ sudo journalctl -l -u ztp.service --no-pager
-- Logs begin at Wed 2016-05-11 16:37:42 UTC, end at Wed 2016-05-11 16:40:39 UTC. --
May 11 16:37:45 cumulus ztp[400]: ztp [400]: /var/lib/cumulus/ztp: Sate Directory does not exist. Creating it...
May 11 16:37:45 cumulus ztp[400]: ztp [400]: /var/run/ztp.lock: Lock File does not exist. Creating it...
May 11 16:37:45 cumulus ztp[400]: ztp [400]: /var/lib/cumulus/ztp/ztp_state.log: State File does not exist. Creating it...
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Looking for ZTP local Script
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell_s6010_s1220-rUNKNOWN
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell_s6010_s1220
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP USB: Looking for unmounted USB devices
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP USB: Parsing partitions
May 11 16:37:45 cumulus ztp[400]: ztp [400]: ZTP USB: Device not found
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Looking for ZTP Script provided by DHCP
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: Attempting to provision via ZTP DHCP from http://192.0.2.1/demo.sh
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: URL response code 200
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Found Marker CUMULUS-AUTOPROVISIONING
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Executing http://192.0.2.1/demo.sh
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: ZTP DHCP: Payload returned code 1
May 11 16:38:45 dell-s6010-01 ztp[400]: ztp [400]: Script returned failure
May 11 16:38:45 dell-s6010-01 systemd[1]: ztp.service: main process exited, code=exited, status=1/FAILURE
May 11 16:38:45 dell-s6010-01 systemd[1]: Unit ztp.service entered failed state.

Instead of running journalctl, you can see the log history by running:

cumulus@switch:~$ cat /var/log/syslog | grep ztp
2016-05-11T16:37:45.132583+00:00 cumulus ztp [400]: /var/lib/cumulus/ztp: State Directory does not exist. Creating it...
2016-05-11T16:37:45.134081+00:00 cumulus ztp [400]: /var/run/ztp.lock: Lock File does not exist. Creating it...
2016-05-11T16:37:45.135360+00:00 cumulus ztp [400]: /var/lib/cumulus/ztp/ztp_state.log: State File does not exist. Creating it...
2016-05-11T16:37:45.185598+00:00 cumulus ztp [400]: ZTP LOCAL: Looking for ZTP local Script
2016-05-11T16:37:45.485084+00:00 cumulus ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell_s6010_s1220-rUNKNOWN
2016-05-11T16:37:45.486394+00:00 cumulus ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell_s6010_s1220
2016-05-11T16:37:45.488385+00:00 cumulus ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64-dell
2016-05-11T16:37:45.489665+00:00 cumulus ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp-x86_64
2016-05-11T16:37:45.490854+00:00 cumulus ztp [400]: ZTP LOCAL: Waterfall search for /var/lib/cumulus/ztp/cumulus-ztp
2016-05-11T16:37:45.492296+00:00 cumulus ztp [400]: ZTP USB: Looking for unmounted USB devices
2016-05-11T16:37:45.493525+00:00 cumulus ztp [400]: ZTP USB: Parsing partitions
2016-05-11T16:37:45.636422+00:00 cumulus ztp [400]: ZTP USB: Device not found
2016-05-11T16:38:43.372857+00:00 cumulus ztp [1805]: Found ZTP DHCP Request
2016-05-11T16:38:45.696562+00:00 cumulus ztp [400]: ZTP DHCP: Looking for ZTP Script provided by DHCP
2016-05-11T16:38:45.698598+00:00 cumulus ztp [400]: Attempting to provision via ZTP DHCP from http://192.0.2.1/demo.sh
2016-05-11T16:38:45.816275+00:00 cumulus ztp [400]: ZTP DHCP: URL response code 200
2016-05-11T16:38:45.817446+00:00 cumulus ztp [400]: ZTP DHCP: Found Marker CUMULUS-AUTOPROVISIONING
2016-05-11T16:38:45.818402+00:00 cumulus ztp [400]: ZTP DHCP: Executing http://192.0.2.1/demo.sh
2016-05-11T16:38:45.834240+00:00 cumulus ztp [400]: ZTP DHCP: Payload returned code 1
2016-05-11T16:38:45.835488+00:00 cumulus ztp [400]: Script returned failure
2016-05-11T16:38:45.876334+00:00 cumulus systemd[1]: ztp.service: main process exited, code=exited, status=1/FAILURE
2016-05-11T16:38:45.879410+00:00 cumulus systemd[1]: Unit ztp.service entered failed state.

If you see that the issue is a script failure, you can modify the script and then run ZTP manually using ztp -v -r <URL/path to that script>, as above.

cumulus@switch:~$ sudo ztp -v -r http://192.0.2.1/demo.sh
Attempting to provision via ZTP Manual from http://192.0.2.1/demo.sh

Broadcast message from root@dell-s6010-01 (ttyS0) (Tue May 10 22:44:17 2019):  

ZTP: Attempting to provision via ZTP Manual from http://192.0.2.1/demo.sh
ZTP Manual: URL response code 200
ZTP Manual: Found Marker CUMULUS-AUTOPROVISIONING
ZTP Manual: Executing http://192.0.2.1/demo.sh
error: ZTP Manual: Payload returned code 1
error: Script returned failure
cumulus@switch:~$ sudo ztp -s
State         enabled
Version       1.0
Result        Script Failure
Date          Mon 20 May 2019 09:31:27 PM UTC
Method        ZTP Manual
URL           http://192.0.2.1/demo.sh

Use the following command to check syslog for information about ZTP:

cumulus@switch:~$ sudo grep -i ztp /var/log/syslog

Common ZTP Script Errors

Could not find referenced script/interpreter in downloaded payload

cumulus@leaf01:~$ sudo cat /var/log/syslog | grep ztp
2018-04-24T15:06:08.887041+00:00 leaf01 ztp [13404]: Attempting to provision via ZTP Manual from http://192.168.0.254/ztp_oob_windows.sh
2018-04-24T15:06:09.106633+00:00 leaf01 ztp [13404]: ZTP Manual: URL response code 200
2018-04-24T15:06:09.107327+00:00 leaf01 ztp [13404]: ZTP Manual: Found Marker CUMULUS-AUTOPROVISIONING
2018-04-24T15:06:09.107635+00:00 leaf01 ztp [13404]: ZTP Manual: Executing http://192.168.0.254/ztp_oob_windows.sh
2018-04-24T15:06:09.132651+00:00 leaf01 ztp [13404]: ZTP Manual: Could not find referenced script/interpreter in downloaded payload.
2018-04-24T15:06:14.135521+00:00 leaf01 ztp [13404]: ZTP Manual: Retrying
2018-04-24T15:06:14.138915+00:00 leaf01 ztp [13404]: ZTP Manual: URL response code 200
2018-04-24T15:06:14.139162+00:00 leaf01 ztp [13404]: ZTP Manual: Found Marker CUMULUS-AUTOPROVISIONING
2018-04-24T15:06:14.139448+00:00 leaf01 ztp [13404]: ZTP Manual: Executing http://192.168.0.254/ztp_oob_windows.sh
2018-04-24T15:06:14.143261+00:00 leaf01 ztp [13404]: ZTP Manual: Could not find referenced script/interpreter in downloaded payload.
2018-04-24T15:06:24.147580+00:00 leaf01 ztp [13404]: ZTP Manual: Retrying
2018-04-24T15:06:24.150945+00:00 leaf01 ztp [13404]: ZTP Manual: URL response code 200
2018-04-24T15:06:24.151177+00:00 leaf01 ztp [13404]: ZTP Manual: Found Marker CUMULUS-AUTOPROVISIONING
2018-04-24T15:06:24.151374+00:00 leaf01 ztp [13404]: ZTP Manual: Executing http://192.168.0.254/ztp_oob_windows.sh
2018-04-24T15:06:24.155026+00:00 leaf01 ztp [13404]: ZTP Manual: Could not find referenced script/interpreter in downloaded payload.
2018-04-24T15:06:39.164957+00:00 leaf01 ztp [13404]: ZTP Manual: Retrying
2018-04-24T15:06:39.165425+00:00 leaf01 ztp [13404]: Script returned failure
2018-04-24T15:06:39.175959+00:00 leaf01 ztp [13404]: ZTP script failed. Exiting...

Errors in syslog for ZTP like those shown above often occur if the script is created (or edited as some point) on a Windows machine. Check to make sure that the \r\n characters are not present in the end-of-line encodings.

Use the cat -v ztp.sh command to view the contents of the script and search for any hidden characters.

root@oob-mgmt-server:/var/www/html# cat -v ./ztp_oob_windows.sh 
#!/bin/bash^M
^M
###################^M
#   ZTP Script^M
###################^M
^M
/usr/cumulus/bin/cl-license -i http://192.168.0.254/license.txt^M
^M
# Clean method of performing a Reboot^M
nohup bash -c 'sleep 2; shutdown now -r "Rebooting to Complete ZTP"' &^M
^M
exit 0^M
^M
# The line below is required to be a valid ZTP script^M
#CUMULUS-AUTOPROVISIONING^M
root@oob-mgmt-server:/var/www/html#

The ^M characters in the output of your ZTP script, as shown above, indicate the presence of Windows end-of-line encodings that you need to remove.

Use the translate (tr) command on any Linux system to remove the '\r' characters from the file.

root@oob-mgmt-server:/var/www/html# tr -d '\r' < ztp_oob_windows.sh > ztp_oob_unix.sh
root@oob-mgmt-server:/var/www/html# cat -v ./ztp_oob_unix.sh 
#!/bin/bash
###################
#   ZTP Script
###################
/usr/cumulus/bin/cl-license -i http://192.168.0.254/license.txt
# Clean method of performing a Reboot
nohup bash -c 'sleep 2; shutdown now -r "Rebooting to Complete ZTP"' &
exit 0
# The line below is required to be a valid ZTP script
#CUMULUS-AUTOPROVISIONING
root@oob-mgmt-server:/var/www/html#

Manually Use the ztp Command

To enable ZTP, use the -e option:

cumulus@switch:~$ sudo ztp -e

Enabling ZTP means that ZTP tries to run the next time the switch boots. However, if ZTP already ran on a previous boot up or if a manual configuration has been found, ZTP will just exit without trying to look for any script.

ZTP checks for these manual configurations during bootup:

  • Password changes
  • Users and groups changes
  • Packages changes
  • Interfaces changes
  • The presence of an installed license

When the switch is booted for the very first time, ZTP records the state of important files that are most likely going to be modified after that the switch is configured. If ZTP is still enabled after a reboot, ZTP compares the recorded state to the current state of these files. If they do not match, ZTP considers that the switch has already been provisioned and exits. These files are only erased after a reset.

To reset ZTP to its original state, use the -R option. This removes the ztp directory and ZTP runs the next time the switch reboots.

cumulus@switch:~$ sudo ztp -R

To disable ZTP, use the -d option:

cumulus@switch:~$ sudo ztp -d

To force provisioning to occur and ignore the status listed in the configuration file, use the -r option:

cumulus@switch:~$ sudo ztp -r cumulus-ztp.sh

To see the current ZTP state, use the -s option:

cumulus@switch:~$ sudo ztp -s
ZTP INFO:
State          disabled
Version        1.0
Result         success
Date           Mon May 20 21:51:04 2019 UTC
Method         Switch manually configured  
URL            None

Notes

System Configuration

This section describes how to configure your Cumulus Linux switch. You can set the date and time, configure authentication, authorization, and accounting and configure access control lists (ACLs), which control the traffic entering your network.

This section also describes the services and daemons that Cumulus Linux uses, and describes how to configure switchd, the daemon at the heart of Cumulus Linux.

An overview of the Network Command Line Utility (NCLU) is also provided.

Network Command Line Utility - NCLU

The Network Command Line Utility (NCLU) is a command line interface for Cumulus Networks products that simplifies the networking configuration process for all users.

NCLU resides in the Linux user space and provides consistent access to networking commands directly through bash, making configuration and troubleshooting simple and easy; no need to edit files or enter modes and sub-modes. NCLU provides these benefits:

The NCLU wrapper utility called net is capable of configuring layer 2 and layer 3 features of the networking stack, installing ACLs and VXLANs, restoring configuration files, as well as providing monitoring and troubleshooting functionality for these features. You can configure both the /etc/network/interfaces and /etc/frr/frr.conf files with net, in addition to running show and clear commands related to ifupdown2 and FRRouting.

NCLU Basics

Use the following workflow to stage and commit changes to Cumulus Linux with NCLU:

  1. Use the net add and net del commands to stage and remove configuration changes.
  2. Use the net pending command to review staged changes.
  3. Use net commit and net abort to commit and delete staged changes.

net commit applies the changes to the relevant configuration files, such as /etc/network/interfaces, then runs necessary follow on commands to enable the configuration, such as ifreload -a.

If two different users try to commit a change at the same time, NCLU displays a warning but implements the change according to the first commit received. The second user will need to abort the commit.

When you have a running configuration, you can review and update the configuration with the following commands:

Tab Completion, Verification, and Inline Help

In addition to tab completion and partial keyword command identification, NCLU includes verification checks to ensure you use the correct syntax. The examples below show the output for incorrect commands:

cumulus@switch:~$ net add bgp router-id 1.1.1.1/32
ERROR: Command not found

Did you mean one of the following?
    net add bgp router-id <ipv4>
        This command is looking for an IP address, not an IP/prefixlen

cumulus@switch:~$ net add bgp router-id 1.1.1.1
cumulus@switch:~$ net add int swp10 mtu <TAB>
    <552-9216> :
cumulus@switch:~$ net add int swp10 mtu 9300
ERROR: Command not found

Did you mean one of the following?
    net add interface <interface> mtu <552-9216>

NCLU has a comprehensive built in help system. In addition to the net man page, you can use ?and help to display available commands:

cumulus@switch:~$ net help

Usage:
    # net <COMMAND> [<ARGS>] [help]
    #
    # net is a command line utility for networking on Cumulus Linux switches.
    #
    # COMMANDS are listed below and have context specific arguments which can
    # be explored by typing "<TAB>" or "help" anytime while using net.
    #
    # Use 'man net' for a more comprehensive overview.

    net abort
    net commit [verbose] [confirm [<number-seconds>]] [description <wildcard>]
    net commit permanent <wildcard>
    net del all
    net help [verbose]
    net pending [json]
    net rollback (<number>|last)
    net rollback description <wildcard-snapshot>
    net show commit (history|<number>|last)
    net show rollback (<number>|last)
    net show rollback description <wildcard-snapshot>
    net show configuration [commands|files|acl|bgp|multicast|ospf|ospf6]
    net show configuration interface [<interface>] [json]

Options:

    # Help commands
    help     : context sensitive information; see section below
    example  : detailed examples of common workflows

    # Configuration commands
    add      : add/modify configuration
    del      : remove configuration


    # Commit buffer commands
    abort    : abandon changes in the commit buffer
    commit   : apply the commit buffer to the system
    pending  : show changes staged in the commit buffer
    rollback : revert to a previous configuration state

    # Status commands
    show     : show command output
    clear    : clear counters, BGP neighbors, etc

cumulus@switch:~$ net help bestpath
The following commands contain keyword(s) 'bestpath'

    net (add|del) bgp bestpath as-path multipath-relax [as-set|no-as-set]
    net (add|del) bgp bestpath compare-routerid
    net (add|del) bgp bestpath med missing-as-worst
    net (add|del) bgp ipv4 labeled-unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp ipv4 unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp ipv6 labeled-unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp ipv6 unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp vrf <text> bestpath as-path multipath-relax [as-set|no-as-set]
    net (add|del) bgp vrf <text> bestpath compare-routerid
    net (add|del) bgp vrf <text> bestpath med missing-as-worst
    net (add|del) bgp vrf <text> ipv4 labeled-unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp vrf <text> ipv4 unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp vrf <text> ipv6 labeled-unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp vrf <text> ipv6 unicast neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net (add|del) bgp vrf <text> neighbor <bgppeer> addpath-tx-bestpath-per-AS
    net add bgp debug bestpath <ip/prefixlen>
    net del bgp debug bestpath [<ip/prefixlen>]
    net show bgp (<ipv4>|<ipv4/prefixlen>|<ipv6>|<ipv6/prefixlen>) [bestpath|multipath] [json]
    net show bgp vrf <text> (<ipv4>|<ipv4/prefixlen>|<ipv6>|<ipv6/prefixlen>) [bestpath|multipath] [json]

You can configure multiple interfaces at once:

cumulus@switch:~$ net add int swp7-9,12,15-17,22 mtu 9216

Add ? (Question Mark) Ability to NCLU

While tab completion is enabled by default, you can also configure NCLU to use the ? (question mark character) to look at available commands. To enable this feature for the cumulus user, open the following file:

cumulus@switch:~$ sudo nano ~/.inputrc

Uncomment the very last line in the .inputrc file so that the file changes from this:

# Uncomment to use ? as an alternative to
# ?: complete

to this:

# Uncomment to use ? as an alternative to
?: complete

Save the file and reconnect to the switch. The ? (question mark) abilitywill work on all subsequent sessions on the switch.

cumulus@switch:~$ net
    abort     :  abandon changes in the commit buffer
    add       :  add/modify configuration
    clear     :  clear counters, BGP neighbors, etc
    commit    :  apply the commit buffer to the system
    del       :  remove configuration
    example   :  detailed examples of common workflows
    help      :  Show this screen and exit
    pending   :  show changes staged in the commit buffer
    rollback  :  revert to a previous configuration state
    show      :  show command output

When the question mark is typed, NCLU autocompletes and shows all available options, but the question mark does not actually appear on the terminal. This is normal, expected behavior.

Built-In Examples

NCLU has a number of built in examples to guide you through basic configuration setup:

cumulus@switch:~$ net example
    acl              :  access-list
    bgp              :  Border Gateway Protocol
    bond             :  bond, port-channel, etc
    bridge           :  a layer2 bridge
    clag             :  Multi-Chassis Link Aggregation
    dhcp             :  Dynamic Host Configuration Protocol
    dot1x            :  Configure, Enable, Delete or Show IEEE 802.1X EAPOL
    evpn             :  Ethernet VPN
    link-settings    :  Physical link parameters
    management-vrf   :  Management VRF
    mlag             :  Multi-Chassis Link Aggregation
    ospf             :  Open Shortest Path First (OSPFv2)
    snmp-server      :  Configure the SNMP server
    syslog           :  Set syslog logging
    vlan-interfaces  :  IP interfaces for VLANs
    voice-vlan       :  VLAN used for IP Phones
    vrr              :  add help text
cumulus@switch:~$ net example bridge

Scenario
========
We are configuring switch1 and would like to configure the following
- configure switch1 as an L2 switch for host-11 and host-12
- enable vlans 10-20
- place host-11 in vlan 10
- place host-12 in vlan 20
- create an SVI interface for vlan 10
- create an SVI interface for vlan 20
- assign IP 10.0.0.1/24 to the SVI for vlan 10
- assign IP 20.0.0.1/24 to the SVI for vlan 20
- configure swp3 as a trunk for vlans 10, 11, 12 and 20
                  swp3
         *switch1 --------- switch2
            /\
      swp1 /  \ swp2
          /    \
         /      \
     host-11   host-12

switch1 net commands
====================
- enable vlans 10-20
switch1# net add vlan 10-20
- place host-11 in vlan 10
- place host-12 in vlan 20
switch1# net add int swp1 bridge access 10
switch1# net add int swp2 bridge access 20
- create an SVI interface for vlan 10
- create an SVI interface for vlan 20
- assign IP 10.0.0.1/24 to the SVI for vlan 10
- assign IP 20.0.0.1/24 to the SVI for vlan 20
switch1# net add vlan 10 ip address 10.0.0.1/24
switch1# net add vlan 20 ip address 20.0.0.1/24
- configure swp3 as a trunk for vlans 10, 11, 12 and 20
switch1# net add int swp3 bridge trunk vlans 10-12,20
switch1# net pending
switch1# net commit

Verification
============
switch1# net show interface
switch1# net show bridge macs

Configure User Accounts

You can configure user accounts in Cumulus Linux with read-only or edit permissions for NCLU:

The examples below demonstrate how to add a new user account or modify an existing user account called myuser.

To add a new user account with NCLU show permissions:

cumulus@switch:~$ sudo adduser --ingroup netshow myuser
Adding user `myuser' ...
Adding new user `myuser' (1001) with group `netshow'...
...

To add NCLU show permissions to a user account that already exists:

cumulus@switch:~$ sudo addgroup myuser netshow
Adding user `myuser' to group `netshow' ...
Adding user myuser to group netshow
Done

To add a new user account with NCLU edit permissions:

cumulus@switch:~$ sudo adduser --ingroup netedit myuser
Adding user `myuser' ...
Adding new user `myuser' (1001) with group `netedit'
...

To add NCLU edit permissions to a user account that already exists:

cumulus@switch:~$ sudo addgroup myuser netedit
Adding user `myuser' to group `netedit' ...
Adding user myuser to group netedit
Done

You can use the adduser command for local user accounts only. You can use the addgroup command for both local and remote user accounts. For a remote user account, you must use the mapping username, such as tacacs3 or radius_user, not the TACACS or RADIUS account name.

If the user tries to run commands that are not allowed, the following error displays:

myuser@switch:~$ net add hostname host01
ERROR: User username does not have permission to make networking changes.

Edit the netd.conf File

Instead of using the NCLU commands described above, you can manually configure users and groups to be able to run NCLU commands.

Edit the /etc/netd.conf file to add users to the users_with_edit and users_with_show lines in the file, then save the file.

For example, if you want the user netoperator to be able to run both edit and show commands, add the user to the users_with_edit and users_with_show lines in the /etc/netd.conf file:

cumulus@switch:~$ sudo nano /etc/netd.conf

# Control which users/groups are allowed to run 'add', 'del',
# 'clear', 'net abort', 'net commit' and restart services
# to apply those changes
users_with_edit = root, cumulus, netoperator
groups_with_edit = root, cumulus

# Control which users/groups are allowed to run 'show' commands
users_with_show = root, cumulus, netoperator
groups_with_show = root, cumulus

To configure a new user group to use NCLU, add that group to the groups_with_edit and groups_with_show lines in the file.

Use caution giving edit permissions to groups. For example, do not give edit permissions to the tacacs group.

Restart the netd Service

Whenever you modify netd.conf or when NSS services change, you must restart the netd service for the changes to take effect:

cumulus@switch:~$ sudo systemctl restart netd.service

Back Up the Configuration to a Single File

You can easily back up your NCLU configuration to a file by outputting the results of net show configuration commands to a file, then retrieving the contents of the file using the source command. You can then view the configuration at any time or copy it to other switches and use the source command to apply that configuration to those switches.

For example, to copy the configuration of a leaf switch called leaf01, run the following command:

cumulus@leaf01:~$ net show configuration commands >> leaf01.txt

With the commands all stored in a single file, you can now copy this file to another ToR switch in your network called leaf01 and apply the configuration by running:

cumulus@leaf01:~$ source leaf01.txt

Advanced Configuration

NCLU needs no initial configuration; however, if you need to modify certain configuration, you must manually update the /etc/netd.conf file. You can configure this file to allow different permission levels for users to edit configurations and run show commands. The file also contains a blacklist that hides less frequently used terms from the tabbed autocomplete.

After you edit the netd.conf file, restart the netd service for the changes to take effect.

cumulus@switch:~$ sudo nano /etc/netd.conf
cumulus@switch:~$ sudo systemctl restart netd.service
Configuration Variable Default Setting Description
show_linux_commandFalseWhen true, displays the Linux command running in the background.
color_diffsTrueWhen true, the diffs shown in net pending and net commit use colors.
enable_<component>TrueWhen true, enables you to configure the component with NCLU. For example, when enable_frr is true, you can use NCLU to configure FRR.
users_with_editroot, cumulusSets the Linux users with root edit privileges.
groups_with_editroot, cumulusSets the Linux groups with root edit privileges.
users_with_showroot, cumulusControls which users are allowed to run show commands.
groups_with_showroot, cumulusControls which groups are allowed to run show commands.
ifupdown_blacklistaddress-purge, bond-ad-actor-sys-prio, bond-ad-actor-system, bond-num-grat-arp,bond-num-unsol-na, bond-use-carrier, bond-xmit-hash-policy, bridge-bridgeprio, bridge-fd, bridge-hashel, bridge-hashmax, bridge-hello, bridge-igmp-querier-src, bridge-maxage, bridge-maxwait, bridge-mclmc, bridge-mclmi bridge-mcmi, bridge-mcqi, bridge-mcqpi, bridge-mcqri, bridge-mcrouter, bridge-mcsqc, bridge-mcsqi, bridge-pathcosts, bridge-port-pvids, bridge-port-vids, bridge-portprios, bridge-waitport, broadcast, link-type, mstpctl-ageing, mstpctl-fdelay, mstpctl-forcevers, mstpctl-hello, mstpctl-maxage, mstpctl-maxhops, mstpctl-portp2p, mstpctl-portpathcost, mstpctl-portrestrtcn, mstpctl-treeportcost, mstpctl-treeportprio, mstpctl-txholdcount, netmask, preferred-lifetime, scope, vxlan-ageing, vxlan-learning, vxlan-port, up, down, bridge-gcint, bridge-mcqifaddr, bridge-mcqv4srcHides corner case command options from tab complete, to simplify and streamline output.

net provides an environment variable to set where the net output is directed. To only use stdout, set the NCLU_TAB_STDOUT environment variable to true. The value is not case sensitive.

Setting Date and Time

Setting the time zone, date and time requires root privileges; use sudo.

Set the Time Zone

You can use one of two methods to set the time zone on the switch:

Edit the /etc/timezone File

To see the current time zone, list the contents of /etc/timezone:

cumulus@switch:~$ cat /etc/timezone
US/Eastern

Edit the file to add your desired time zone. A list of valid time zones can be found here.

Use the following command to apply the new time zone immediately.

cumulus@switch:~$ sudo dpkg-reconfigure --frontend noninteractive tzdata

Use the Guided Wizard

To set the time zone using the guided wizard, run dpkg-reconfigure tzdata as root:

cumulus@switch:~$ sudo dpkg-reconfigure tzdata

For more information, see the Debian System Administrator's Manual - Time.

Set the Date and Time

The switch contains a battery backed hardware clock that maintains the time while the switch is powered off and in between reboots. When the switch is running, the Cumulus Linux operating system maintains its own software clock.

During boot up, the time from the hardware clock is copied into the operating system’s software clock. The software clock is then used for all timekeeping responsibilities. During system shutdown, the software clock is copied back to the battery backed hardware clock.

You can set the date and time on the software clock using the date command. First, determine your current time zone:

cumulus@switch:~$ date +%Z

If you need to reconfigure the current time zone, refer to the instructions above.

Then, to set the system clock according to the time zone configured:

cumulus@switch:~$ sudo date -s "Tue Jan 12 00:37:13 2016"

See man date(1) for more information.

You can write the current value of the system (software) clock to the hardware clock using the hwclock command:

cumulus@switch:~$ sudo hwclock -w

See man hwclock(8) for more information.

Use NTP

The ntpd daemon running on the switch implements the NTP protocol. It synchronizes the system time with time servers listed in the /etc/ntp.conf file. The ntpd daemon is started at boot by default. See man ntpd(8) for details.

If you intend to run this service within a VRF, including the management VRF, follow these steps for configuring the service.

Configure NTP Servers

The default NTP configuration comprises the following servers, which are listed in the /etc/ntpd.conf file:

To add the NTP server or servers you want to use:

NCLU Commands

Run the following commands. Include the iburst option to increase the sync speed.

cumulus@switch:~$ net add time ntp server 4.cumulusnetworks.pool.ntp.org iburst
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands add the NTP server to the list of servers in the /etc/ntp.conf file:

# pool.ntp.org maps to about 1000 low-stratum NTP servers.  Your server will
# pick a different set every time it starts up.  Please consider joining the
# pool: <http://www.pool.ntp.org/join.html>
server 0.cumulusnetworks.pool.ntp.org iburst
server 1.cumulusnetworks.pool.ntp.org iburst
server 2.cumulusnetworks.pool.ntp.org iburst
server 3.cumulusnetworks.pool.ntp.org iburst
server 4.cumulusnetworks.pool.ntp.org iburst
Linux Commands

Edit the /etc/ntp.conf file to add or update NTP server information:

cumulus@switch:~$ sudo nano /etc/ntp.conf
# pool.ntp.org maps to about 1000 low-stratum NTP servers.  Your server will
# pick a different set every time it starts up.  Please consider joining the
# pool: <http://www.pool.ntp.org/join.html>
server 0.cumulusnetworks.pool.ntp.org iburst
server 1.cumulusnetworks.pool.ntp.org iburst
server 2.cumulusnetworks.pool.ntp.org iburst
server 3.cumulusnetworks.pool.ntp.org iburst
server 4.cumulusnetworks.pool.ntp.org iburst

To set the initial date and time with NTP before starting the ntpd daemon, run the ntpd -q command. This command is the same as ntpdate, which is to be retired and no longer available.

Be aware that ntpd -q can hang if the time servers are not reachable.

To verify that ntpd is running on the system:

cumulus@switch:~$ ps -ef | grep ntp
ntp       4074     1  0 Jun20 ?        00:00:33 /usr/sbin/ntpd -p /var/run/ntpd.pid -g -u 101:102

To check the NTP peer status:

NCLU Commands

Run the net show time ntp servers command:

cumulus@switch:~$ net show time ntp servers 
      remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
+minime.fdf.net  58.180.158.150   3 u  140 1024  377   55.659    0.339   1.464
+69.195.159.158  128.138.140.44   2 u  259 1024  377   41.587    1.011   1.677
*chl.la          216.218.192.202  2 u  210 1024  377    4.008    1.277   1.628
+vps3.drown.org  17.253.2.125     2 u  743 1024  377   39.319   -0.316   1.384
Linux Commands

Run the ntpq -p command:

cumulus@switch:~$ ntpq -p
      remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
+ec2-34-225-6-20 129.6.15.30      2 u   73 1024  377   70.414   -2.414   4.110
+lax1.m-d.net    132.163.96.1     2 u   69 1024  377   11.676    0.155   2.736
*69.195.159.158  199.102.46.72    2 u  133 1024  377   48.047   -0.457   1.856
-2.time.dbsinet. 198.60.22.240    2 u 1057 1024  377   63.973    2.182   2.692

To remove one or more NTP servers:

NCLU Commands

Run the net del time ntp <server> iburst command. The following example commands remove some of the default NTP servers.

cumulus@switch:~$ net del time ntp server 0.cumulusnetworks.pool.ntp.org iburst
cumulus@switch:~$ net del time ntp server 1.cumulusnetworks.pool.ntp.org iburst
cumulus@switch:~$ net del time ntp server 2.cumulusnetworks.pool.ntp.org iburst
cumulus@switch:~$ net del time ntp server 3.cumulusnetworks.pool.ntp.org iburst
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/ntp.conf file to delete the NTP servers.

cumulus@switch:~$ sudo nano /etc/ntp.conf
...
# pool.ntp.org maps to about 1000 low-stratum NTP servers.  Your server will
# pick a different set every time it starts up.  Please consider joining the
# pool: <http://www.pool.ntp.org/join.html>
server 4.cumulusnetworks.pool.ntp.org iburst
...

Specify the NTP Source Interface

By default, the source interface that NTP uses is eth0. To change the source interface:

NCLU Commands

Run the net add time ntp source <interface> command. The following command example changes the NTP source interface to swp10.

cumulus@switch:~$ net add time ntp source swp10
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration snippet in the ntp.conf file:

...
# Specify interfaces
interface listen swp10
...
Linux Commands

Edit the /etc/ntp.conf file and modify the entry under the # Specify interfaces comment. The following example shows that the NTP source interface is swp10.

cumulus@switch:~$ sudo nano /etc/ntp.conf
...
# Specify interfaces
interface listen swp10
...

Use NTP in a DHCP Environment

You can use DHCP to specify your NTP servers. Ensure that the DHCP-generated configuration file named /run/ntp.conf.dhcp exists. This file is generated by the /etc/dhcp/dhclient-exit-hooks.d/ntp script and is a copy of the default /etc/ntp.conf with a modified server list from the DHCP server. If this file does not exist and you plan on using DHCP in the future, you can copy your current /etc/ntp.conf file to the location of the DHCP file.

To use DHCP to specify your NTP servers, run the sudo -E systemctl edit ntp.service command and add the ExecStart= line:

cumulus@switch:~$ sudo -E systemctl edit ntp.service
[Service]
ExecStart=
ExecStart=/usr/sbin/ntpd -n -u ntp:ntp -g -c /run/ntp.conf.dhcp

The sudo -E systemctl edit ntp.service command always updates the base ntp.service even if ntp@mgmt.service is used. The ntp@mgmt.service is re-generated automatically.

To validate that your configuration, run these commands:

cumulus@switch:~$ sudo systemctl restart ntp
cumulus@switch:~$ sudo systemctl status -n0 ntp.service

If the state is not Active, or the alternate configuration file does not appear in the ntp command line, it is likely that a mistake was made. In this case, correct the mistake and rerun the three commands above to verify.

When you use the above procedure to specify your NTP servers, the NCLU commands for changing NTP settings do not take effect.

Configure NTP with Authorization Keys

For added security, you can configure NTP to use authorization keys.

Configure the NTP server:

  1. Create a .keys file, such as /etc/ntp.keys. Specify a key identifier (a number from 1-65535), an encryption method (M for MD5), and the password. The following provides an example:
#
# PLEASE DO NOT USE THE DEFAULT VALUES HERE.
#
#65535  M  akey
#1      M  pass

1  M  CumulusLinux!
  1. In the /etc/ntp/ntp.conf file, add a pointer to the /etc/ntp.keys file you created above and specify the key identifier. For example:
keys /etc/ntp/ntp.keys
trustedkey 1
controlkey 1
requestkey 1
  1. Restart NTP with the sudo systemctl restart ntp command.

Configure the NTP client (the Cumulus Linux switch):

  1. Create the same .keys file you created on the NTP server (/etc/ntp.keys). For example:
cumulus@switch:~$  sudo nano /etc/ntp.keys
#
# PLEASE DO NOT USE THE DEFAULT VALUES HERE.
#
#65535  M  akey
#1      M  pass

1  M  CumulusLinux!
  1. Edit the /etc/ntp.conf file to specify the server you want to use, the key identifier, and a pointer to the /etc/ntp.keys file you created in step 1. For example:
cumulus@switch:~$ sudo nano /etc/ntp.conf
...
# You do need to talk to an NTP server or two (or three).
#pool ntp.your-provider.example
# OR
#server ntp.your-provider.example

# pool.ntp.org maps to about 1000 low-stratum NTP servers.  Your server will
# pick a different set every time it starts up.  Please consider joining the
# pool: <http://www.pool.ntp.org/join.html>
#server 0.cumulusnetworks.pool.ntp.org iburst
#server 1.cumulusnetworks.pool.ntp.org iburst
#server 2.cumulusnetworks.pool.ntp.org iburst
#server 3.cumulusnetworks.pool.ntp.org iburst
server 10.50.23.121 key 1

#keys
keys /etc/ntp.keys
trustedkey 1
controlkey 1
requestkey 1
...
  1. Restart NTP in the active VRF (default or management). For example:
cumulus@switch:~$ systemctl restart ntp@mgmt.service
  1. Wait a few minutes, then run the ntpq -c as command to verify the configuration:
cumulus@switch:~$ ntpq -c as

ind assid status  conf reach auth condition  last_event cnt
===========================================================
  1 40828  f014   yes   yes   ok     reject   reachable  1

After authorization is accepted, you see the following command output:

cumulus@switch:~$ ntpq -c as

ind assid status  conf reach auth condition  last_event cnt
===========================================================
  1 40828  f61a   yes   yes   ok   sys.peer    sys_peer  1

Precision Time Protocol (PTP) Boundary Clock

With the growth of low latency and high performance applications, precision timing has become increasingly important. Precision Time Protocol (PTP) is used to synchronize clocks in a network and is capable of sub-microsecond accuracy. The clocks are organized in a master-slave hierarchy. The slaves are synchronized to their masters, which can be slaves to their own masters. The hierarchy is created and updated automatically by the best master clock (BMC) algorithm, which runs on every clock. The grandmaster clock is the top-level master and is typically synchronized by using a Global Positioning System (GPS) time source to provide a high-degree of accuracy.

A boundary clock has multiple ports; one or more master ports and one or more slave ports. The master ports provide time (the time can originate from other masters further up the hierarchy) and the slave ports receive time. The boundary clock absorbs sync messages in the slave port, uses that port to set its clock, then generates new sync messages from this clock out of all of its master ports.

Cumulus Linux includes the linuxptp package for PTP, which uses the phc2sys daemon to synchronize the PTP clock with the system clock.

  • Cumulus Linux currently supports PTP on the Mellanox Spectrum ASIC only.
  • PTP is supported in boundary clock mode only (the switch provides timing to downstream servers; it is a slave to a higher-level clock and a master to downstream clocks).
  • The switch uses hardware time stamping to capture timestamps from an Ethernet frame at the physical layer. This allows PTP to account for delays in message transfer and greatly improves the accuracy of time synchronization.
  • Only IPv4/UDP PTP packets are supported.
  • Only a single PTP domain per network is supported. A PTP domain is a network or a portion of a network within which all the clocks are synchronized.

In the following example, boundary clock 2 receives time from Master 1 (the grandmaster) on a PTP slave port, sets its clock and passes the time down from the PTP master port to boundary clock 1. Boundary clock 1 receives the time on a PTP slave port, sets its clock and passes the time down the hierarchy through the PTP master ports to the hosts that receive the time.

Enable the PTP Boundary Clock on the Switch

To enable the PTP boundary clock on the switch:

  1. Open the /etc/cumulus/switchd.conf file in a text editor and add the following line:
ptp.timestamping = TRUE
  1. Restart switchd:
cumulus@switch:~$ sudo systemctl restart switchd.service

Configure the PTP Boundary Clock

To configure a boundary clock:

  1. Configure the interfaces on the switch that you want to use for PTP. Each interface must be configured as a layer 3 routed interface with an IP address.

    • PTP is supported on BGP unnumbered interfaces.
    • PTP is not supported on switched virtual interfaces (SVIs).

cumulus@switch:~$ net add interface swp13s0 ip address 10.0.0.9/32
cumulus@switch:~$ net add interface swp13s1 ip address 10.0.0.10/32
  1. Configure PTP options on the switch:

    • Set the gm-capable option to no to configure the switch to be a boundary clock.
    • Set the priority, which selects the best master clock. You can set priority 1 or 2. For each priority, you can use a number between 0 and 255. The default priority is 255. For the boundary clock, use a number above 128. The lower priority is applied first.
    • Add the time-stamping parameter. The switch automatically enables hardware time-stamping to capture timestamps from an Ethernet frame at the physical layer. If you are testing PTP in a virtual environment, hardware time-stamping is not available; however the time-stamping parameter is still required.
    • Add the PTP master and slave interfaces. You do not specify which is a master interface and which is a slave interface; this is determined by the PTP packet received. The following commands provide an example configuration:
cumulus@switch:~$ net add ptp global gm-capable no
cumulus@switch:~$ net add ptp global priority2 254
cumulus@switch:~$ net add ptp global priority1 254
cumulus@switch:~$ net add ptp global time-stamping
cumulus@switch:~$ net add ptp interface swp13s0
cumulus@switch:~$ net add ptp interface swp13s1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The ptp4l man page describes all the configuration parameters.

  1. Restart the ptp4l and phc2sys daemons:
cumulus@switch:~$ sudo systemctl restart ptp4l.service phc2sys.service

The configuration is saved in the /etc/ptp4l.conf file.

  1. Enable the services to start at boot time:
cumulus@switch:~$ sudo systemctl enable ptp4l.service phc2sys.service

Example Configuration

In the following example, the boundary clock on the switch receives time from Master 1 (the grandmaster) on PTP slave port swp3s0, sets its clock and passes the time down through PTP master ports swp3s1, swp3s2, and swp3s3 to the hosts that receive the time.

The configuration for the above example is shown below. The example assumes that you have already configured the layer 3 routed interfaces (swp3s0, swp3s1, swp3s2, and swp3s3) you want to use for PTP.

cumulus@switch:~$ net add ptp global gm-capable no
cumulus@switch:~$ net add ptp global priority2 254
cumulus@switch:~$ net add ptp global priority1 254
cumulus@switch:~$ net add ptp global time-stamping
cumulus@switch:~$ net add ptp interface swp3s0
cumulus@switch:~$ net add ptp interface swp3s1
cumulus@switch:~$ net add ptp interface swp3s2
cumulus@switch:~$ net add ptp interface swp3s3
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Verify PTP Boundary Clock Configuration

To view a summary of the PTP configuration on the switch, run the net show configuration ptp command:

cumulus@switch:~$ net show configuration ptp
ptp
  global

    slaveOnly
      0

    priority1
      255

    priority2
      255

    domainNumber
      0

    logging_level
      5

    path_trace_enabled
      0

    use_syslog
      1

    verbose
      0

    summary_interval
      0

    time_stamping
      hardware

    gmCapable
      0
  swp15s0
  swp15s1
...

View PTP Status Information

To view PTP status information, run the net show ptp parent_data_set command:

cumulus@switch:~$ net show ptp parent_data_set
parent_data_set
===============
parentPortIdentity                    000200.fffe.000001-1
parentStats                           0
observedParentOffsetScaledLogVariance 0xffff
observedParentClockPhaseChangeRate    0x7fffffff
grandmasterPriority1                  127
gm.ClockClass                         248
gm.ClockAccuracy                      0xfe
gm.OffsetScaledLogVariance            0xffff
grandmasterPriority2                  127
grandmasterIdentity                   000200.fffe.000001

To view the additional PTP status information, including the delta in nanoseconds from the master clock, run the sudo pmc -u -b 0 'GET TIME_STATUS_NP' command:

cumulus@switch:~$ sudo pmc -u -b 0 'GET TIME_STATUS_NP'
sending: GET TIME_STATUS_NP
    7cfe90.fffe.f56dfc-0 seq 0 RESPONSE MANAGEMENT TIME_STATUS_NP
        master_offset              12610
        ingress_time               1525717806521177336
        cumulativeScaledRateOffset +0.000000000
        scaledLastGmPhaseChange    0
        gmTimeBaseIndicator        0
        lastGmPhaseChange          0x0000'0000000000000000.0000
        gmPresent                  true
        gmIdentity                 000200.fffe.000005
    000200.fffe.000005-1 seq 0 RESPONSE MANAGEMENT TIME_STATUS_NP
        master_offset              0
        ingress_time               0
        cumulativeScaledRateOffset +0.000000000
        scaledLastGmPhaseChange    0
        gmTimeBaseIndicator        0
        lastGmPhaseChange          0x0000'0000000000000000.0000
        gmPresent                  false
        gmIdentity                 000200.fffe.000005
    000200.fffe.000006-1 seq 0 RESPONSE MANAGEMENT TIME_STATUS_NP
        master_offset              5544033534
        ingress_time               1525717812106811842
        cumulativeScaledRateOffset +0.000000000
        scaledLastGmPhaseChange    0
        gmTimeBaseIndicator        0
        lastGmPhaseChange          0x0000'0000000000000000.0000
        gmPresent                  true
        gmIdentity                 000200.fffe.000005

Delete PTP Boundary Clock Configuration

To delete PTP configuration, delete the PTP master and slave interfaces. The following example commands delete the PTP interfaces swp3s0, swp3s1, and swp3s2.

cumulus@switch:~$ net del ptp interface swp3s0
cumulus@switch:~$ net del ptp interface swp3s1
cumulus@switch:~$ net del ptp interface swp3s2
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Authentication, Authorization and Accounting

This section descibes how to set up user accounts, ssh for remote access, LDAP authentication, TACACS Plus, and RADIUS AAA.

SSH for Remote Access

You can generate authentication keys to access a Cumulus Linux switch securely with the ssh-keygen component of the Secure Shell (SSH) protocol. Cumulus Linux uses the OpenSSH package to provide this functionality. This section describes how to generate an SSH key pair.

Generate an SSH Key Pair

  1. To generate the SSH key pair, run the ssh-keygen command and follow the prompts:

    To configure a completely passwordless system, do not enter a passphrase when prompted in the following step.

cumulus@leaf01:~$ ssh-keygen
Generating public/private rsa key pair.
Enter file in which to save the key (/home/cumulus/.ssh/id_rsa):
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in /home/cumulus/.ssh/id_rsa.
Your public key has been saved in /home/cumulus/.ssh/id_rsa.pub.
The key fingerprint is:
5a:b4:16:a0:f9:14:6b:51:f6:f6:c0:76:1a:35:2b:bb cumulus@leaf04
The key's randomart image is:
+---[RSA 2048]----+
|      +.o   o    |
|     o * o . o   |
|    o + o O o    |
|     + . = O     |
|      . S o .    |
|       +   .     |
|      .   E      |
|                 |
|                 |
+-----------------+
  1. To copy the generated public key to the desired location, run the ssh-copy-id command and follow the prompts:
cumulus@leaf01:~$ ssh-copy-id -i /home/cumulus/.ssh/id_rsa.pub cumulus@leaf02
The authenticity of host 'leaf02 (192.168.0.11)' can't be established.
ECDSA key fingerprint is b1:ce:b7:6a:20:f4:06:3a:09:3c:d9:42:de:99:66:6e.
Are you sure you want to continue connecting (yes/no)? yes
/usr/bin/ssh-copy-id: INFO: attempting to log in with the new key(s), to filter out any that are already installed
/usr/bin/ssh-copy-id: INFO: 1 key(s) remain to be installed -- if you are prompted now it is to install the new keys
cumulus@leaf01's password:

Number of key(s) added: 1

ssh-copy-id does not work if the username on the remote switch is different from the username on the local switch. To work around this issue, use the scp command instead:

cumulus@leaf01:~$ scp .ssh/id_rsa.pub cumulus@leaf02:.ssh/authorized_keys
Enter passphrase for key '/home/cumulus/.ssh/id_rsa':
id_rsa.pub
  1. Connect to the remote switch to confirm that the authentication keys are in place:
cumulus@leaf01:~$ ssh cumulus@leaf02

Welcome to Cumulus VX (TM)

Cumulus VX (TM) is a community supported virtual appliance designed for
experiencing, testing and prototyping Cumulus Networks' latest technology.
For any questions or technical support, visit our community site at:
http://community.cumulusnetworks.com

The registered trademark Linux (R) is used pursuant to a sublicense from LMI,
the exclusive licensee of Linus Torvalds, owner of the mark on a world-wide basis.
Last login: Thu Sep 29 16:56:54 2016

User Accounts

By default, Cumulus Linux has two user accounts: cumulus and root.

The cumulus account:

The root account:

For optimal security, change the default password with the passwd command before you configure Cumulus Linux on the switch.

You can add additional user accounts as needed. Like the cumulus account, these accounts must use sudo to execute privileged commands; be sure to include them in the sudo group. For example:

cumulus@switch:~$ sudo adduser NEWUSERNAME sudo

To access the switch without a password, you need to boot into a single shell/user mode.

You can add and configure user accounts in Cumulus Linux with read-only or edit permissions for NCLU. For more information, see Configure User Accounts.

Enable Remote Access for the root User

The root user does not have a password and cannot log into a switch using SSH. This default account behavior is consistent with Debian. To connect to a switch using the root account, you can do one of the following:

Generate an SSH Key for the root Account

  1. In a terminal on your host system (not the switch), check to see if a key already exists:
root@host:~# ls -al ~/.ssh/

The name of the key is similar to id_dsa.pub, id_rsa.pub, or id_ecdsa.pub.

  1. If a key does not exist, generate a new one by first creating the RSA key pair:
root@host:~# ssh-keygen -t rsa
  1. You are prompted to enter a file in which to save the key (/root/.ssh/id_rsa). Press Enter to use the home directory of the root user or provide a different destination.

  2. You are prompted to enter a passphrase (empty for no passphrase). This is optional but it does provide an extra layer of security.

  3. The public key is now located in /root/.ssh/id_rsa.pub. The private key (identification) is now located in /root/.ssh/id_rsa.

  4. Copy the public key to the switch. SSH to the switch as the cumulus user, then run:

cumulus@switch:~$ sudo mkdir -p /root/.ssh
cumulus@switch:~$ echo <SSH public key string> | sudo tee -a /root/.ssh/authorized_keys

Set the root User Password

  1. Run the following command:
cumulus@switch:~$ sudo passwd root
  1. Change the PermitRootLogin setting in the /etc/ssh/sshd_config file from without-password to yes.
cumulus@switch:~$ sudo nano /etc/ssh/sshd_config
...
# Authentication:
LoginGraceTime 120
PermitRootLogin yes
StrictModes yes
...  
  1. Restart the ssh service:
cumulus@switch:~$ sudo systemctl reload ssh.service

Using sudo to Delegate Privileges

By default, Cumulus Linux has two user accounts: root and cumulus. The cumulus account is a normal user and is in the group sudo.

You can add more user accounts as needed. Like the cumulus account, these accounts must use sudo to execute privileged commands.

sudo Basics

sudo allows you to execute a command as superuser or another user as specified by the security policy. See man sudo(8) for details.

The default security policy is sudoers, which is configured using /etc/sudoers. Use /etc/sudoers.d/ to add to the default sudoers policy. See man sudoers(5) for details.

Use visudo only to edit the sudoers file; do not use another editor like vi or emacs. See man visudo(8) for details.

When creating a new file in /etc/sudoers.d, use visudo -f. This option performs sanity checks before writing the file to avoid errors that prevent sudo from working.

Errors in the sudoers file can result in losing the ability to elevate privileges to root. You can fix this issue only by power cycling the switch and booting into single user mode. Before modifying sudoers, enable the root user by setting a password for the root user.

By default, users in the sudo group can use sudo to execute privileged commands. To add users to the sudo group, use the useradd(8) or usermod(8) command. To see which users belong to the sudo group, see /etc/group (man group(5)).

You can run any command as sudo, including su. A password is required.

The example below shows how to use sudo as a non-privileged user cumulus to bring up an interface:

cumulus@switch:~$ ip link show dev swp1
3: swp1: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast master br0 state DOWN mode DEFAULT qlen 500
link/ether 44:38:39:00:27:9f brd ff:ff:ff:ff:ff:ff

cumulus@switch:~$ ip link set dev swp1 up
RTNETLINK answers: Operation not permitted

cumulus@switch:~$ sudo ip link set dev swp1 up
Password:

umulus@switch:~$ ip link show dev swp1
3: swp1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master br0 state UP mode DEFAULT qlen 500
link/ether 44:38:39:00:27:9f brd ff:ff:ff:ff:ff:ff

sudoers Examples

The following examples show how you grant as few privileges as necessary to a user or group of users to allow them to perform the required task. For each example, the system group noc is used; groups are prefixed with an %.

When executed by an unprivileged user, the example commands below must be prefixed with sudo.

CategoryPrivilegeExample Commandsudoers Entry
MonitoringSwitch port informationethtool -m swp1%noc ALL=(ALL) NOPASSWD:/sbin/ethtool
MonitoringSystem diagnosticscl-support%noc ALL=(ALL) NOPASSWD:/usr/cumulus/bin/cl-support
MonitoringRouting diagnosticscl-resource-query%noc ALL=(ALL) NOPASSWD:/usr/cumulus/bin/cl-resource-query
Image managementInstall imagesonie-select http://lab/install.bin%noc ALL=(ALL) NOPASSWD:/usr/cumulus/bin/onie-select
Package managementAny apt-get commandapt-get update or apt-get install%noc ALL=(ALL) NOPASSWD:/usr/bin/apt-get
Package managementJust apt-get updateapt-get update%noc ALL=(ALL) NOPASSWD:/usr/bin/apt-get update
Package managementInstall packagesapt-get install vim%noc ALL=(ALL) NOPASSWD:/usr/bin/apt-get install *
Package managementUpgradingapt-get upgrade%noc ALL=(ALL) NOPASSWD:/usr/bin/apt-get upgrade
NetfilterInstall ACL policiescl-acltool -i%noc ALL=(ALL) NOPASSWD:/usr/cumulus/bin/cl-acltool
NetfilterList iptables rulesiptables -L%noc ALL=(ALL) NOPASSWD:/sbin/iptables
L1 + 2 featuresAny LLDP commandlldpcli show neighbors / configure%noc ALL=(ALL) NOPASSWD:/usr/sbin/lldpcli
L1 + 2 featuresJust show neighborslldpcli show neighbors%noc ALL=(ALL) NOPASSWD:/usr/sbin/lldpcli show neighbors*
InterfacesModify any interfaceip link set dev swp1 {updown}
InterfacesUp any interfaceifup swp1%noc ALL=(ALL) NOPASSWD:/sbin/ifup
InterfacesDown any interfaceifdown swp1%noc ALL=(ALL) NOPASSWD:/sbin/ifdown
InterfacesUp/down only swp2ifup swp2 / ifdown swp2%noc ALL=(ALL) NOPASSWD:/sbin/ifup swp2,/sbin/ifdown swp2
InterfacesAny IP address changeip addr {adddel} 192.0.2.1/30 dev swp1
InterfacesOnly set IP addressip addr add 192.0.2.1/30 dev swp1%noc ALL=(ALL) NOPASSWD:/sbin/ip addr add *
Ethernet bridgingAny bridge commandbrctl addbr br0 / brctl delif br0 swp1%noc ALL=(ALL) NOPASSWD:/sbin/brctl
Ethernet bridgingAdd bridges and interfacesbrctl addbr br0 / brctl addif br0 swp1%noc ALL=(ALL) NOPASSWD:/sbin/brctl addbr *,/sbin/brctl addif *
Spanning treeSet STP propertiesmstpctl setmaxage br2 20%noc ALL=(ALL) NOPASSWD:/sbin/mstpctl
TroubleshootingRestart switchdsystemctl restart switchd.service%noc ALL=(ALL) NOPASSWD:/usr/sbin/service switchd *
TroubleshootingRestart any servicesystemctl cron switchd.service%noc ALL=(ALL) NOPASSWD:/usr/sbin/service
TroubleshootingPacket capturetcpdump%noc ALL=(ALL) NOPASSWD:/usr/sbin/tcpdump
L3Add static routesip route add 10.2.0.0/16 via 10.0.0.1%noc ALL=(ALL) NOPASSWD:/bin/ip route add *
L3Delete static routesip route del 10.2.0.0/16 via 10.0.0.1%noc ALL=(ALL) NOPASSWD:/bin/ip route del *
L3Any static route changeip route *%noc ALL=(ALL) NOPASSWD:/bin/ip route *
L3Any iproute commandip *%noc ALL=(ALL) NOPASSWD:/bin/ip
L3Non-modal OSPFcl-ospf area 0.0.0.1 range 10.0.0.0/24%noc ALL=(ALL) NOPASSWD:/usr/bin/cl-ospf

LDAP Authentication and Authorization

Cumulus Linux uses Pluggable Authentication Modules (PAM) and Name Service Switch (NSS) for user authentication. NSS enables PAM to use LDAP to provide user authentication, group mapping, and information for other services on the system.

There are three common ways to configure LDAP authentication on Linux: you can use libnss-ldap, libnss-ldapd, or libnss-sss. This chapter describes libnss-ldapd only. From internal testing, this library worked best with Cumulus Linux and is the easiest to configure, automate, and troubleshoot.

Install libnss-ldapd

The libldap-2.4-2 and libldap-common LDAP packages are already installed on the Cumulus Linux image; however you need to install these additional packages to use LDAP authentication:

To install the additional packages, run the following command:

cumulus@switch:~$ sudo apt-get install libnss-ldapd libpam-ldapd ldap-utils nslcd

You can also install these packages even if the switch is not connected to the internet, as they are contained in the cumulus-local-apt-archive repository that is embedded in the Cumulus Linux disk image.

Follow the interactive prompts to specify the LDAP URI, search base distinguished name (DN), and services that must have LDAP lookups enabled. You need to select at least the passwd, group, and shadow services (press space to select a service). When done, click OK. This creates a very basic LDAP configuration using anonymous bind and initiates user search under the base DN specified.

After the dialog closes, the install process prints information similar to the following:

/etc/nsswitch.conf: enable LDAP lookups for group
/etc/nsswitch.conf: enable LDAP lookups for passwd
/etc/nsswitch.conf: enable LDAP lookups for shadow

After the installation is complete, the name service caching daemon (nslcd) runs. This service handles all the LDAP protocol interactions and caches information returned from the LDAP server. ldap is appended in the /etc/nsswitch.conf file, as is the secondary information source for passwd, group, and shadow. The local files (/etc/passwd, /etc/groups and /etc/shadow) are used first, as specified by the compat source.

passwd: compat ldap
group: compat ldap
shadow: compat ldap

Cumulus Networks recommends that you keep compat as the first source in NSS for passwd, group, and shadow. This prevents you from getting locked out of the system.

Entering incorrect information during the installation process might produce configuration errors. You can correct the information after installation by editing certain configuration files.

Be sure to restart netd after editing the files.

Alternative Installation Method Using debconf-utils

Instead of running the installer and following the interactive prompts, as described above, you can pre-seed the installer parameters using debconf-utils.

  1. Run apt-get install debconf-utils and create the pre-seeded parameters using debconf-set-selections. Provide the appropriate answers.
  2. Run debconf-show <pkg> to check the settings. Here is an example of how to pre-seed answers to the installer questions using debconf-set-selections:
root# debconf-set-selections <<'zzzEndOfFilezzz'

# LDAP database user. Leave blank will be populated later!

nslcd nslcd/ldap-binddn  string

# LDAP user password. Leave blank!
nslcd nslcd/ldap-bindpw   password

# LDAP server search base:
nslcd nslcd/ldap-base string  ou=support,dc=rtp,dc=example,dc=test

# LDAP server URI. Using ldap over ssl.
nslcd nslcd/ldap-uris string  ldaps://myadserver.rtp.example.test

# New to 0.9. restart cron, exim and others libraries without asking
nslcd libraries/restart-without-asking: boolean true

# LDAP authentication to use:
# Choices: none, simple, SASL
# Using simple because its easy to configure. Security comes by using LDAP over SSL
# keep /etc/nslcd.conf 'rw' to root for basic security of bindDN password
nslcd nslcd/ldap-auth-type    select  simple

# Don't set starttls to true
nslcd nslcd/ldap-starttls     boolean false

# Check server's SSL certificate:
# Choices: never, allow, try, demand
nslcd nslcd/ldap-reqcert      select  never

# Choices: Ccreds credential caching - password saving, Unix authentication, LDAP Authentication , Create home directory on first time login, Ccreds credential caching - password checking
# This is where "mkhomedir" pam config is activated that allows automatic creation of home directory
libpam-runtime        libpam-runtime/profiles multiselect     ccreds-save, unix, ldap, mkhomedir , ccreds-check

# for internal use; can be preseeded
man-db        man-db/auto-update      boolean true

# Name services to configure:
# Choices: aliases, ethers, group, hosts, netgroup, networks, passwd, protocols, rpc, services,  shadow
libnss-ldapd  libnss-ldapd/nsswitch   multiselect     group, passwd, shadow
libnss-ldapd  libnss-ldapd/clean_nsswitch     boolean false

## define platform specific libnss-ldapd debconf questions/answers. 
## For demo used amd64.
libnss-ldapd:amd64    libnss-ldapd/nsswitch   multiselect     group, passwd, shadow
libnss-ldapd:amd64    libnss-ldapd/clean_nsswitch     boolean false
# libnss-ldapd:powerpc   libnss-ldapd/nsswitch   multiselect     group, passwd, shadow
# libnss-ldapd:powerpc    libnss-ldapd/clean_nsswitch     boolean false

zzzEndOfFilezzz

Update the nslcd.conf File

After installation, update the main configuration file (/etc/nslcd.conf) to accommodate the expected LDAP server settings.

This section documents some of the more important options that relate to security and how queries are handled. For details on all the available configuration options, read the nslcd.conf man page.

After first editing the /etc/nslcd.conf file and/or enabling LDAP in the /etc/nsswitch.conf file, you must restart netd with the sudo systemctl restart netd command. If you disable LDAP, you need to restart the netd service.

Connection

The LDAP client starts a session by connecting to the LDAP server on TCP and UDP port 389 or on port 636 for LDAPS. Depending on the configuration, this connection might be unauthenticated (anonymous bind); otherwise, the client must provide a bind user and password. The variables used to define the connection to the LDAP server are the URI and bind credentials.

The URI is mandatory and specifies the LDAP server location using the FQDN or IP address. The URI also designates whether to use ldap:// for clear text transport, or ldaps:// for SSL/TLS encrypted transport. You can also specify an alternate port in the URI. In production environments, the LDAPS protocol is recommended so that all communications are secure.

After the connection to the server is complete, the BIND operation authenticates the session. The BIND credentials are optional, and if not specified, an anonymous bind is assumed. This is typically not allowed in most production environments. Configure authenticated (Simple) BIND by specifying the user (binddn) and password (bindpw) in the configuration. Another option is to use SASL (Simple Authentication and Security Layer) BIND, which provides authentication services using other mechanisms, like Kerberos. Contact your LDAP server administrator for this information as it depends on the configuration of the LDAP server and the credentials that are created for the client device.

# The location at which the LDAP server(s) should be reachable.
uri ldaps://ldap.example.com
# The DN to bind with for normal lookups.
binddn cn=CLswitch,ou=infra,dc=example,dc=com
bindpw CuMuLuS

Search Function

When an LDAP client requests information about a resource, it must connect and bind to the server. Then, it performs one or more resource queries depending on the lookup. All search queries sent to the LDAP server are created using the configured search base, filter, and the desired entry (uid=myuser) being searched. If the LDAP directory is large, this search might take a significant amount of time. It is a good idea to define a more specific search base for the common maps (passwd and group).

# The search base that will be used for all queries.
base dc=example,dc=com
# Mapped search bases to speed up common queries.
base passwd ou=people,dc=example,dc=com
base group ou=groups,dc=example,dc=com

Search Filters

It is also common to use search filters to specify criteria used when searching for objects within the directory. This is used to limit the search scope when authenticating users. The default filters applied are:

filter passwd (objectClass=posixAccount)
filter group (objectClass=posixGroup)

Attribute Mapping

The map configuration allows you to override the attributes pushed from LDAP. To override an attribute for a given map, specify the attribute name and the new value. This is useful to ensure that the shell is bash and the home directory is /home/cumulus:

map    passwd homeDirectory "/home/cumulus"
map    passwd shell "/bin/bash"

In LDAP, the map refers to one of the supported maps specified in the manpage for nslcd.conf (such as passwd or group).

Create Home Directory on Login

If you want to use unique home directories, run the sudo pam-auth-update command and select Create home directory on login in the PAM configuration dialog (press the space bar to select the option). Select OK, then press Enter to save the update and close the dialog.

cumulus@switch:~$ sudo pam-auth-update

The home directory for any user that logs in (using LDAP or not) is created and populated with the standard dotfiles from /etc/skel if it does not already exist.

When nslcd starts, you might see an error message similar to the following (where 5816 is the nslcd PID):

nslcd[5816]: unable to dlopen /usr/lib/x86_64-linux-gnu/sasl2/libsasldb.so: libdb-5.3.so: cannot open
shared object file: No such file or directory

You can safely ignore this message. The libdb package and resulting log messages from nslcd do not cause any issues when you use LDAP as a client for login and authentication.

Example Configuration

Here is an example configuration using Cumulus Linux.

# /etc/nslcd.conf
# nslcd configuration file. See nslcd.conf(5)
# for details.

# The user and group nslcd should run as.
uid nslcd
gid nslcd

# The location at which the LDAP server(s) should be reachable.
uri ldaps://myadserver.rtp.example.test

# The search base that will be used for all queries.
base ou=support,dc=rtp,dc=example,dc=test

# The LDAP protocol version to use.
#ldap_version 3

# The DN to bind with for normal lookups.
# defconf-set-selections doesn't seem to set this. so have to manually set this.
binddn CN=cumulus admin,CN=Users,DC=rtp,DC=example,DC=test
bindpw 1Q2w3e4r!

# The DN used for password modifications by root.
#rootpwmoddn cn=admin,dc=example,dc=com

# SSL options
#ssl off (default)
# Not good does not prevent man in the middle attacks
#tls_reqcert demand(default)
tls_cacertfile /etc/ssl/certs/rtp-example-ca.crt

# The search scope.
#scope sub

# Add nested group support
# Supported in nslcd 0.9 and higher.
# default wheezy install of nslcd supports on 0.8. wheezy-backports has 0.9
nss_nested_groups yes

# Mappings for Active Directory
# (replace the SIDs in the objectSid mappings with the value for your domain)
# "dsquery * -filter (samaccountname=testuser1) -attr ObjectSID" where cn == 'testuser1'
pagesize 1000
referrals off
idle_timelimit 1000

# Do not allow uids lower than 100 to login (aka Administrator)
# not needed as pam already has this support
# nss_min_uid 1000

# This filter says to get all users who are part of the cumuluslnxadm group. Supports nested groups.
# Example, mary is part of the snrnetworkadm group which is part of cumuluslnxadm group
# Ref: http://msdn.microsoft.com/en-us/library/aa746475%28VS.85%29.aspx (LDAP_MATCHING_RULE_IN_CHAIN)
filter passwd (&(Objectclass=user)(!(objectClass=computer))(memberOf:1.2.840.113556.1.4.1941:=cn=cumuluslnxadm,ou=groups,ou=support,dc=rtp,dc=example,dc=test))
map    passwd uid           sAMAccountName
map    passwd uidNumber     objectSid:S-1-5-21-1391733952-3059161487-1245441232
map    passwd gidNumber     objectSid:S-1-5-21-1391733952-3059161487-1245441232
map    passwd homeDirectory "/home/$sAMAccountName"
map    passwd gecos         displayName
map    passwd loginShell    "/bin/bash"

# Filter for any AD group or user in the baseDN. the reason for filtering for the
# user to make sure group listing for user files don't say '<user> <gid>'. instead will say '<user> <user>'
# So for cosmetic reasons..nothing more.
filter group (&(|(objectClass=group)(Objectclass=user))(!(objectClass=computer)))
map    group gidNumber     objectSid:S-1-5-21-1391733952-3059161487-1245441232
map    group cn            sAMAccountName

Troubleshooting

nslcd Debug Mode

When setting up LDAP authentication for the first time, Cumulus Networks recommends you turn off the nslcd service using the systemctl stop nslcd.service command (or the systemctl stop nslcd@mgmt.service if you are running the service in a management VRF) and run it in debug mode. Debug mode works whether you are using LDAP over SSL (port 636) or an unencrypted LDAP connection (port 389).

cumulus@switch:~$ sudo systemctl stop nslcd.service
cumulus@switch:~$ sudo nslcd -d

After you enable debug mode, run the following command to test LDAP queries:

cumulus@switch:~$ getent passwd

If LDAP is configured correctly, the following messages appear after you run the getent command:

nslcd: DEBUG: accept() failed (ignored): Resource temporarily unavailable
nslcd: [8e1f29] DEBUG: connection from pid=11766 uid=0 gid=0
nslcd: [8e1f29] <passwd(all)> DEBUG: myldap_search(base="dc=example,dc=com", filter="(objectClass=posixAccount)")
nslcd: [8e1f29] <passwd(all)> DEBUG: ldap_result(): uid=myuser,ou=people,dc=example,dc=com
nslcd: [8e1f29] <passwd(all)> DEBUG: ldap_result(): ... 152 more results
nslcd: [8e1f29] <passwd(all)> DEBUG: ldap_result(): end of results (162 total)

In the output above, <passwd(all)> indicates that the entire directory structure is queried.

You can query a specific user with the following command:

cumulus@switch:~$ getent passwd myuser

You can replace myuser with any username on the switch. The following debug output indicates that user myuser exists:

nslcd: DEBUG: add_uri(ldap://10.50.21.101)
nslcd: version 0.8.10 starting
nslcd: DEBUG: unlink() of /var/run/nslcd/socket failed (ignored): No such file or directory
nslcd: DEBUG: setgroups(0,NULL) done
nslcd: DEBUG: setgid(110) done
nslcd: DEBUG: setuid(107) done
nslcd: accepting connections
nslcd: DEBUG: accept() failed (ignored): Resource temporarily unavailable
nslcd: [8b4567] DEBUG: connection from pid=11369 uid=0 gid=0
nslcd: [8b4567] <passwd="myuser"> DEBUG: myldap_search(base="dc=cumulusnetworks,dc=com", filter="(&(objectClass=posixAccount)(uid=myuser))")
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_initialize(ldap://<ip_address>)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_rebind_proc()
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_PROTOCOL_VERSION,3)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_DEREF,0)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_TIMELIMIT,0)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_TIMEOUT,0)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_NETWORK_TIMEOUT,0)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_REFERRALS,LDAP_OPT_ON)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_set_option(LDAP_OPT_RESTART,LDAP_OPT_ON)
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_simple_bind_s(NULL,NULL) (uri="ldap://<ip_address>")
nslcd: [8b4567] <passwd="myuser"> DEBUG: ldap_result(): end of results (0 total)

Common Problems

SSL/TLS

NSCD

nscd --invalidate = passwd
nscd --invalidate = group
cumulus@switch:~$ sudo nscd -K
cumulus@switch:~$ sudo systemctl restart nslcd.service

If you are running the nslcd service in a management VRF, you need to run the systemctl restart nslcd@mgmt.service command instead of the systemctl restart nslcd.service command. For example:

cumulus@switch:~$ sudo nscd -K
cumulus@switch:~$ sudo systemctl restart nslcd@mgmt.service

LDAP

# ldapsearch -D 'cn=CLadmin' -w 'CuMuLuS' "(&(ObjectClass=inetOrgUser)(uid=myuser))"
# /etc/nsswitch.conf
passwd:       ldap compat

Configure LDAP Authorization

Linux uses the sudo command to allow non-administrator users (such as the default cumulus user account) to perform privileged operations. To control the users authorized to use sudo, the /etc/sudoers file and files located in the /etc/sudoers.d/ directory define a series of rules. Typically, the rules are based on groups, but can also be defined for specific users. You can add sudo rules using the group names from LDAP. For example, if a group of users are associated with the group netadmin, you can add a rule to give those users sudo privileges. Refer to the sudoers manual (man sudoers) for a complete usage description. The following shows an example in the /etc/sudoers file:

# The basic structure of a user specification is "who where = (as_whom) what ".
%sudo ALL=(ALL:ALL) ALL
%netadmin ALL=(ALL:ALL) ALL

Active Directory Configuration

Active Directory (AD) is a fully featured LDAP-based NIS server create by Microsoft. It offers unique features that classic OpenLDAP servers do not have. AD can be more complicated to configure on the client and each version works a little differently with Linux-based LDAP clients. Some more advanced configuration examples, from testing LDAP clients on Cumulus Linux with Active Directory (AD/LDAP), are available in our knowledge base.

LDAP Verification Tools

Typically, password and group information is retrieved from LDAP and cached by the LDAP client daemon. To test the LDAP interaction, you can use these command-line tools to trigger an LDAP query from the device. This helps to create the best filters and verify the information sent back from the LDAP server.

Identify a User with the id Command

The id command performs a username lookup by following the lookup information sources in NSS for the passwd service. This simply returns the user ID, group ID and the group list retrieved from the information source. In the following example, the user cumulus is locally defined in /etc/passwd, and myuser is on LDAP. The NSS configuration has the passwd map configured with the sources compat ldap:

cumulus@switch:~$ id cumulus
uid=1000(cumulus) gid=1000(cumulus) groups=1000(cumulus),24(cdrom),25(floppy),27(sudo),29(audio),30(dip),44(video),46(plugdev)
cumulus@switch:~$ id myuser
uid=1230(myuser) gid=3000(Development) groups=3000(Development),500(Employees),27(sudo)

getent

The getent command retrieves all records found with NSS for a given map. It can also retrieve a specific entry under that map. You can perform tests with the passwd, group, shadow, or any other map configured in the /etc/nsswitch.conf file. The output from this command is formatted according to the map requested. For the passwd service, the structure of the output is the same as the entries in /etc/passwd. The group map outputs the same structure as /etc/group.

In this example, looking up a specific user in the passwd map, the user cumulus is locally defined in /etc/passwd, and myuser is only in LDAP.

cumulus@switch:~$ getent passwd cumulus
cumulus:x:1000:1000::/home/cumulus:/bin/bash
cumulus@switch:~$ getent passwd myuser
myuser:x:1230:3000:My Test User:/home/myuser:/bin/bash

In the next example, looking up a specific group in the group service, the group cumulus is locally defined in /etc/groups, and netadmin is on LDAP.

cumulus@switch:~$ getent group cumulus
cumulus:x:1000:
cumulus@switch:~$ getent group netadmin
netadmin:*:502:larry,moe,curly,shemp

Running the command getent passwd or getent group without a specific request returns all local and LDAP entries for the passwd and group maps.

The ldapsearch command performs LDAP operations directly on the LDAP server. This does not interact with NSS. This command helps display what the LDAP daemon process is receiving back from the server. The command has many options. The simplest option uses anonymous bind to the host and specifies the search DN and the attribute to look up.

cumulus@switch:~$ ldapsearch -H ldap://ldap.example.com -b dc=example,dc=com -x uid=myuser
Click to expand the command output ...
# extended LDIF
#
# LDAPv3
# base <dc=example,dc=com> with scope subtree
# filter: uid=myuser
# requesting: ALL
#
# myuser, people, example.com
dn: uid=myuser,ou=people,dc=example,dc=com
cn: My User
displayName: My User
gecos: myuser
gidNumber: 3000
givenName: My
homeDirectory: /home/myuser
initials: MU
loginShell: /bin/bash
mail: myuser@example.com
objectClass: inetOrgPerson
objectClass: posixAccount
objectClass: shadowAccount
objectClass: top
shadowExpire: -1
shadowFlag: 0
shadowMax: 999999
shadowMin: 8
shadowWarning: 7
sn: User
uid: myuser
uidNumber: 1234

# search result
search: 2
result: 0 Success

# numResponses: 2
# numEntries: 1

sdsds

To use NCLU, a user must be in either the netshow or netedit NCLU group in the LDAP database. You can either:

In the following example, a user that is not in the netshow or netedit NCLU group in the LDAP database runs the NCLU net show version command, which produces an error:

cumulus@switch:~$ net show version
ERROR: 'getpwuid(): uid not found: 1126'
See /var/log/netd.log for more details

To add the the user to the netshow or netedit NCLU group in the LDAP database, either edit the /etc/group file manually or use the sudo adduser USERNAME netshow command, then restart netd. For example, to add the user bill to the netshow group:

cumulus@switch:~$ sudo adduser bill netshow
Adding user `bill' to group `netshow' ...
Adding user bill to group netshow
Done.

cumulus@switch:~$ sudo systemctl restart netd

Now, the user can run the NCLU net show commands successfully:

cumulus@switch:~$ net show version
NCLU_VERSION=1.0-cl4u1~1555625956.7cfe305
DISTRIB_ID="Cumulus Linux"
DISTRIB_RELEASE=4.0.0~1555370771.772c26b6
DISTRIB_DESCRIPTION="Cumulus Linux 4.0.0~1555370771.772c26b6"

LDAP Browsers

There are several GUI LDAP clients available that help you work with LDAP servers. These are free tools that show the structure of the LDAP database graphically.

TACACS+

Cumulus Linux implements TACACS+ client AAA (Accounting, Authentication, and Authorization) in a transparent way with minimal configuration. The client implements the TACACS+ protocol as described in this IETF document. There is no need to create accounts or directories on the switch. Accounting records are sent to all configured TACACS+ servers by default. Use of per-command authorization requires additional setup on the switch.

Supported Features

Install the TACACS+ Client Packages

You can install the TACACS+ packages even if the switch is not connected to the internet, as they are contained in the cumulus-local-apt-archive repository that is embedded in the Cumulus Linux disk image.

To install all required packages, run these commands:

cumulus@switch:~$ sudo -E apt-get update
cumulus@switch:~$ sudo -E apt-get install tacplus-client

Configure the TACACS+ Client

After installing TACACS+, edit the /etc/tacplus_servers file to add at least one server and one shared secret (key). You can specify the server and secret parameters in any order anywhere in the file. Whitespace (spaces or tabs) are not allowed. For example, if your TACACS+ server IP address is 192.168.0.30 and your shared secret is tacacskey, add these parameters to the /etc/tacplus_servers file:

secret=tacacskey
server=192.168.0.30

Cumulus Linux supports a maximum of seven TACACS+ servers. To specify multiple servers, add one per line to the /etc/tacplus_servers file.

Connections are made in the order in which they are listed in this file. In most cases, you do not need to change any other parameters. You can add parameters used by any of the packages to this file, which affects all the TACACS+ client software. For example, the timeout value for NSS lookups (see description below) is set to 5 seconds by default in the /etc/tacplus_nss.conf file, whereas the timeout value for other packages is 10 seconds and is set in the /etc/tacplus_servers file. The timeout value is per connection to the TACACS+ servers. (If authorization is configured per command, the timeout occurs for each command.) There are several (typically four) connections to the server per login attempt from PAM, as well as two or more through NSS. Therefore, with the default timeout values, a TACACS+ server that is not reachable can delay logins by a minute or more per unreachable server. If you must list unreachable TACACS+ servers, place them at the end of the server list and consider reducing the timeout values.

When you add or remove TACACS+ servers, you must restart auditd (with the systemctl restart auditd command) or you must send a signal (with killall -HUP audisp-tacplus) before audisp-tacplus rereads the configuration to see the changed server list.

You can also configure the IP address used as the source IP address when communicating with the TACACS+ server. See TACACS Configuration Parameters below for the full list of TACACS+ parameters.

Following is the complete list of the TACACS+ client configuration files, and their use.

FilenameDescription
/etc/tacplus_serversThis is the primary file that requires configuration after installation. The file is used by all packages with include=/etc/tacplus_servers parameters in the other configuration files that are installed. Typically, this file contains the shared secrets; make sure that the Linux file mode is 600.
/etc/nsswitch.confWhen the libnss_tacplus package is installed, this file is configured to enable tacplus lookups via libnss_tacplus. If you replace this file by automation or other means, you need to add tacplus as the first lookup method for the passwd database line.
/etc/tacplus_nss.confThis file sets the basic parameters for libnss_tacplus. It includes a debug variable for debugging NSS lookups separately from other client packages.
/usr/share/pam-configs/tacplusThis is the configuration file for pam-auth-update to generate the files in the next row. These configurations are used at login, by su, and by ssh.
/etc/pam.d/common-*The /etc/pam.d/common-* files are updated for tacplus authentication. The files are updated with pam-auth-update, when libpam-tacplus is installed or removed.
/etc/sudoers.d/tacplusThis file allows TACACS+ privilege level 15 users to run commands with sudo. The file includes an example (commented out) of how to enable privilege level 15 TACACS users to use sudo without having to enter a password and provides an example of how to enable all TACACS users to run specific commands with sudo. Only edit this file with the visudo -f /etc/sudoers.d/tacplus command.
audisp-tacplus.confThis is the audisp plugin configuration file. Typically, no modifications are required.
/etc/audisp/audisp-tac_plus.confThis is the TACACS+ server configuration file for accounting. Typically, no modifications are required. You can use this configuration file when you only want to debug TACACS+ accounting issues, not all TACACS+ users.
/etc/audit/rules.d/audisp-tacplus.rulesThe auditd rules for TACACS+ accounting. The augenrules command uses all rule files to generate the rules file (described below).
/etc/audit/audit.rulesThis is the audit rules file generated when auditd is installed.

You can edit the /etc/pam.d/common-* files manually. However, if you run pam-auth-update again after making the changes, the update fails. Only perform configuration in /usr/share/pam-configs/tacplus, then run pam-auth-update.

TACACS+ Authentication (login)

The initial authentication configuration is done through the PAM modules and an updated version of the libpam-tacplus package. When the package is installed, the PAM configuration is updated in /etc/pam.d with the pam-auth-update command. If you have made changes to your PAM configuration, you need to integrate these changes yourself. If you are also using LDAP with the libpam-ldap package, you might need to edit the PAM configuration to ensure the LDAP and TACACS ordering that you prefer. The libpam-tacplus are configured to skip over rules and the values in the success=2 might require adjustments to skip over LDAP rules.

A user privilege level is determined by the TACACS+ privilege attribute priv_lvl for the user that is returned by the TACACS+ server during the user authorization exchange. The client accepts the attribute in either the mandatory or optional forms and also accepts priv-lvl as the attribute name. The attribute value must be a numeric string in the range 0 to 15, with 15 the most privileged level.

By default, TACACS+ users at privilege levels other than 15 are not allowed to run sudo commands and are limited to commands that can be run with standard Linux user permissions.

TACACS+ Client Sequencing

Due to SSH and login processing mechanisms, Cumulus Linux needs to know the following very early in the AAA sequence:

The only way to do this for non-local users — that is, users not present in the local password file — is to send a TACACS+ authorization request as the first communication with the TACACS+ server, prior to the authentication and before a password is requested from the user logging in.

Some TACACS+ servers need special configuration to allow authorization requests prior to authentication. Contact Cumulus Support if your TACACS+ server does not allow the initial authorization request.

Local Fallback Authentication

If a site wants to allow local fallback authentication for a user when none of the TACACS servers can be reached you can add a privileged user account as a local account on the switch.

To configure local fallback authentication:

  1. Edit the /etc/nsswitch.conf file to remove the keyword tacplus from the line starting with passwd. (You need to add the keyword back in step 3.)

    An example of the /etc/nsswitch.conf file with the keyword tacplus removed from the line starting with passwd is shown below.

cumulus@switch:~$ sudo vi /etc/nsswitch.conf
#
# Example configuration of GNU Name Service Switch functionality.
# If you have the `glibc-doc-reference' and `info' packages installed, try:
# `info libc "Name Service Switch"' for information about this file.

passwd:         files
group:          tacplus files
shadow:         files
gshadow:        files
...
  1. To enable the local privileged user to run sudo and NCLU commands, run the adduser commands shown below. In the example commands, the TACACS account name is tacadmin.

    The first adduser command prompts for information and a password. You can skip most of the requested information by pressing ENTER.

cumulus@switch:~$ sudo adduser --ingroup tacacs tacadmin
cumulus@switch:~$ sudo adduser tacadmin netedit
cumulus@switch:~$ sudo adduser tacadmin sudo
  1. Edit the /etc/nsswitch.conf file to add the keyword tacplus back to the line starting with passwd (the keyword you removed in the first step).

  2. Restart the netd service with the following command:

cumulus@switch:~$ sudo systemctl restart netd

TACACS+ Accounting

TACACS+ accounting is implemented with the audisp module, with an additional plugin for auditd/audisp. The plugin maps the auid in the accounting record to a TACACS login, based on the auid and sessionid. The audisp module requires libnss_tacplus and uses the libtacplus_map.so library interfaces as part of the modified lipam_tacplus` package.

Communication with the TACACS+ servers is done with the libsimple-tacact1 library, through dlopen(). A maximum of 240 bytes of command name and arguments are sent in the accounting record, due to the TACACS+ field length limitation of 255 bytes.

All Linux commands result in an accounting record, including commands run as part of the login process or as sub-processes of other commands. This can sometimes generate a large number of accounting records.

Configure the IP address and encryption key of the server in the /etc/tacplus_servers file. Minimal configuration to auditd and audisp is necessary to enable the audit records necessary for accounting. These records are installed as part of the package.

audisp-tacplus installs the audit rules for command accounting. Modifying the configuration files is not usually necessary. However, when a management VRF is configured, the accounting configuration does need special modification because the auditd service starts prior to networking. It is necessary to add the vrf parameter and to signal the audisp-tacplus process to reread the configuration. The example below shows that the management VRF is named mgmt. You can place the vrf parameter in either the /etc/tacplus_servers file or in the /etc/audisp/audisp-tac_plus.conf file.

vrf=mgmt

After editing the configuration file, send the HUP signal killall -HUP audisp-tacplus to notify the accounting process to reread the file.

All sudo commands run by TACACS+ users generate accounting records against the original TACACS+ login name.

For more information, refer to the audisp.8 and auditd.8 man pages.

Configure NCLU for TACACS+ Users

When you install or upgrade TACACS+ packages, mapped user accounts are created automatically. All tacacs0 through tacacs15 users are added to the netshow group.

For any TACACS+ users to execute net add, net del, and net commit commands and to restart services with NCLU, you need to add those users to the users_with_edit variable in the /etc/netd.conf file. Cumulus Networks recommends you add the tacacs15 user and, depending upon your policies, other users (tacacs1 through tacacs14) to this variable.

To give a TACACS+ user access to the show commands, add the tacacs group to the groups_with_show variable.

Do not add the tacacs group to the groups_with_edit variable; this is dangerous and can potentially enable any user to log into the switch as the root user.

To add the users, edit the /etc/netd.conf file:

cumulus@switch:~$ sudo nano /etc/netd.conf
...
# Control which users/groups are allowed to run "add", "del",
# "clear", "abort", and "commit" commands.
users_with_edit = root, cumulus, tacacs15
groups_with_edit = netedit

# Control which users/groups are allowed to run "show" commands
users_with_show = root, cumulus
groups_with_show = netshow, netedit, tacacs
...

After you save and exit the netd.conf file, restart the netd service. Run:

cumulus@switch:~$ sudo systemctl restart netd

TACACS+ Per-command Authorization

The tacplus-auth command handles the per-command authorization. To make this an enforced authorization, you must change the TACACS+ login to use a restricted shell, with a very limited executable search path. Otherwise, the user can bypass the authorization. The tacplus-restrict utility simplifies the setup of the restricted environment. The example below initializes the environment for the tacacs0 user account. This is the account used for TACACS+ users at privilege level 0.

tacuser0@switch:~$ sudo tacplus-restrict -i -u tacacs0 -a command1 command2 ... commandN

If the user/command combination is not authorized by the TACACS+ server, a message similar to the following displays:

tacuser0@switch:~$ net show version
net not authorized by TACACS+ with given arguments, not executing

The following table provides the command options:

OptionDescription
-iInitializes the environment. You only need to issue this option once per username.
-aYou can invoke the utility with the -a option as many times as desired. For each command in the -a list, a symbolic link is created from tacplus-auth to the relative portion of the command name in the local bin subdirectory. You also need to enable these commands on the TACACS+ server (refer to the TACACS+ server documentation). It is common to have the server allow some options to a command, but not others.
-fRe-initializes the environment. If you need to restart, issue the -f option with -i to force the re-initialization; otherwise, repeated use of -i is ignored.
As part of the initialization:
- The user’s shell is changed to /bin/rbash.
- Any existing dot files are saved.
- A limited environment is set up that does not allow general command execution, but instead allows only commands from the user’s local bin subdirectory.

For example, if you want to allow the user to be able to run the net and ip commands (if authorized by the TACACS+ server), use the command:

cumulus@switch:~$ sudo tacplus-restrict -i -u tacacs0 -a ip net

After running this command, examine the tacacs0 directory::

cumulus@switch:~$ sudo ls -lR ~tacacs0
total 12
lrwxrwxrwx 1 root root 22 Nov 21 22:07 ip -> /usr/sbin/tacplus-auth
lrwxrwxrwx 1 root root 22 Nov 21 22:07 net -> /usr/sbin/tacplus-auth

Other than shell built-ins, the only two commands the privilege level 0 TACACS users can run are the ip and net commands.

If you mistakenly add potential commands with the -a option, you can remove them. The example below shows how to remove the net command:

cumulus@switch:~$ sudo rm ~tacacs0/bin/net

You can remove all commands as follows:

cumulus@switch:~$ sudo rm ~tacacs0/bin/*

Use the man command on the switch for more information on tacplus-auth and tacplus-restrict.

cumulus@switch:~$ man tacplus-auth tacplus-restrict

NSS Plugin

When used with pam_tacplus, TACACS+ authenticated users can log in without a local account on the system using the NSS plugin that comes with the tacplus_nss package. The plugin uses the mapped tacplus information if the user is not found in the local password file, provides the getpwnam() and getpwuid()entry point,s and uses the TACACS+ authentication functions.

The plugin asks the TACACS+ server if the user is known, and then for relevant attributes to determine the privilege level of the user. When the libnss_tacplus package is installed, nsswitch.conf is modified to set tacplus as the first lookup method for passwd. If the order is changed, lookups return the local accounts, such as tacacs0

If the user is not found, a mapped lookup is performed using the libtacplus.so exported functions. The privilege level is appended to tacacs and the lookup searches for the name in the local password file. For example, privilege level 15 searches for the tacacs15 user. If the user is found, the password structure is filled in with information for the user.

If the user is not found, the privilege level is decremented and checked again until privilege level 0 (user tacacs0) is reached. This allows use of only the two local users tacacs0 and tacacs15, if minimal configuration is desired.

TACACS Configuration Parameters

The recognized configuration options are the same as the libpam_tacplus command line arguments; however, not all pam_tacplus options are supported. These configuration parameters are documented in the tacplus_servers.5 man page, which is part of the libpam-tacplus package.

The table below describes the configuration options available:

Configuration OptionDescription
debugThe output debugging information through syslog(3).
Note: Debugging is heavy, including passwords. Do not leave debugging enabled on a production switch after you have completed troubleshooting.
secret=STRINGThe secret key used to encrypt and decrypt packets sent to and received from the server.
You can specify the secret key more than once in any order with respect to the server= parameter. When fewer secret= parameters are specified, the last secret given is used for the remaining servers.
Only use this parameter in files such as /etc/tacplus_servers that are not world readable.
server=hostname
server=ip-address
Adds a TACACS+ server to the servers list. Servers are queried in turn until a match is found, or no servers remain in the list. Can be specified up to 7 times. An IP address can be optionally followed by a port number, preceded by a “:". The default port is 49.
Note: When sending accounting records, the record is sent to all servers in the list if acct_all=1, which is the default.
source_ip=ipv4-addressSets the IP address used as the source IP address when communicating with the TACACS+ server. You must specify an IPv4 address. IPv6 addresses and hostnames are not supported. The address must must be valid for the interface being used.
timeout=secondsTACACS+ server(s) communication timeout.
This parameter defaults to 10 seconds in the /etc/tacplus_servers file, but defaults to 5 seconds in the /etc/tacplus_nss.conf file.
include=/file/nameA supplemental configuration file to avoid duplicating configuration information. You can include up to 8 more configuration files.
min_uid=valueThe minimum user ID that the NSS plugin looks up. Setting it to 0 means uid 0 (root) is never looked up, which is desirable for performance reasons. The value should not be greater than the local TACACS+ user IDs (0 through 15), to ensure they can be looked up.
exclude_users=user1,user2,…A comma-separated list of usernames that are never looked up by the NSS plugin, set in the tacplus_nss.conf file. You cannot use * (asterisk) as a wild card in the list. While it’s not a legal username, bash may lookup this as a user name during pathname completion, so it is included in this list as a username string.
Note: Cumulus Networks strongly recommends that you do not remove the cumulus user from the exclude_users list, because doing so can make it impossible to log in as the cumulus user, which is the primary administrative account in Cumulus Linux. If you do remove the cumulus user, Cumulus Networks recommends you add some other local fallback user that does not rely on TACACS but is a member of sudo and netedit groups, so that these accounts can run sudo and NCLU commands.
login=stringTACACS+ authentication service (pap, chap, or login).
The default value is pap.
user_homedir=1This is not enabled by default. When enabled, a separate home directory for each TACACS+ user is created when the TACACS+ user first logs in. By default, the home directory in the mapping accounts in /etc/passwd (/home/tacacs0 … /home/tacacs15) is used. If the home directory does not exist, it is created with the mkhomedir_helper program, in the same way as pam_mkhomedir.
This option is not honored for accounts with restricted shells when per-command authorization is enabled.
acct_all=1Configuration option for audisp_tacplus and pam_tacplus sending accounting records to all supplied servers (1), or the first server to respond (0).
The default value is 1.
timeout=secondsSets the timeout in seconds for connections to each TACACS+ server.
The default is 10 seconds for all lookups except that NSS lookups use a 5 second timeout.
vrf=vrf-nameIf the management network is in a VRF, set this variable to the VRF name. This is typically mgmt. When this variable is set, the connection to the TACACS+ accounting servers is made through the named VRF.
serviceTACACS+ accounting and authorization service. Examples include shell, pap, raccess, ppp, and slip.
The default value is shell.
protocolTACACS+ protocol field. This option is use dependent. PAM uses the SSH protocol.

Remove the TACACS+ Client Packages

To remove all of the TACACS+ client packages, use the following commands:

cumulus@switch:~$ sudo -E apt-get remove tacplus-client
cumulus@switch:~$ sudo -E apt-get autoremove

To remove the TACACS+ client configuration files as well as the packages (recommended), use this command:

cumulus@switch:~$ sudo -E apt-get autoremove --purge

Troubleshooting

Basic Server Connectivity or NSS Issues

You can use the getent command to determine if TACACS+ is configured correctly and if the local password is stored in the configuration files. In the example commands below, the cumulus user represents the local user, while cumulusTAC represents the TACACS user.

To look up the username within all NSS methods:

cumulus@switch:~$ sudo getent passwd cumulusTAC
cumulusTAC:x:1016:1001:TACACS+ mapped user at privilege level 15,,,:/home/tacacs15:/bin/bash

To look up the user within the local database only:

cumulus@switch:~$ sudo getent -s compat passwd cumulus
cumulus:x:1000:1000:cumulus,,,:/home/cumulus:/bin/bash

To look up the user within the TACACS+ database only:

cumulus@switch:~$ sudo getent -s tacplus passwd cumulusTAC
cumulusTAC:x:1016:1001:TACACS+ mapped user at privilege level 15,,,:/home/tacacs15:/bin/bash

If TACACS does not appear to be working correctly, debug the following configuration files by adding the debug=1 parameter to one or more of these files:

You can also add debug=1 to individual pam_tacplus lines in /etc/pam.d/common*.

All log messages are stored in /var/log/syslog.

Incorrect Shared Key

The TACACS client on the switch and the TACACS server should have the same shared secret key. If this key is incorrect, the following message is printed to syslog:

2017-09-05T19:57:00.356520+00:00 leaf01 sshd[3176]: nss_tacplus: TACACS+ server 192.168.0.254:49 read failed with protocol error (incorrect shared secret?) user cumulus

Issues with Per-command Authorization

To debug TACACS user command authorization, have the TACACS+ user enter the following command at a shell prompt, then try the command again:

tacuser0@switch:~$ export TACACSAUTHDEBUG=1

When this debugging is enabled, additional information is shown for the command authorization conversation with the TACACS+ server:

tacuser0@switch:~$ net pending
tacplus-auth: found matching command (/usr/bin/net) request authorization
tacplus-auth: error connecting to 10.0.3.195:49 to request authorization for net: Transport endpoint is not connected
tacplus-auth: cmd not authorized (16)
tacplus-auth: net not authorized from 192.168.3.189:49
net not authorized by TACACS+ with given arguments, not executing
tacuser0@switch:~$ net show version
tacplus-auth: found matching command (/usr/bin/net) request authorization
tacplus-auth: error connecting to 10.0.3.195:49 to request authorization for net: Transport endpoint is not connected
tacplus-auth: 192.168.3.189:49 authorized command net
tacplus-auth: net authorized, executing
DISTRIB_ID="Cumulus Linux"
DISTRIB_RELEASE=4.0.0
DISTRIB_DESCRIPTION="Cumulus Linux 4.0.0"

To disable debugging:

tacuser0@switch:~$ export -n TACACSAUTHDEBUG

Debug Issues with Accounting Records

If you have added or deleted TACACS+ servers from the configuration files, make sure you notify the audisp plugin with this command:

cumulus@switch:~$ sudo killall -HUP audisp-tacplus

If accounting records are still not being sent, add debug=1 to the /etc/audisp/audisp-tac_plus.conf file, then issue the command above to notify the plugin. Ask the TACACS+ user to run a command and examine the end of /var/log/syslog for messages from the plugin. You can also check the auditing log file /var/log/audit audit.log to be sure the auditing records are being written. If they are not, restart the audit daemon with:

cumulus@switch:~$ sudo systemctl restart auditd.service

TACACS Component Software Descriptions

The following table describes the different pieces of software involved with delivering TACACS.

Package NameDescription
audisp-tacplus_1.0.0-1-cl3u3This package uses auditing data from auditd to send accounting records to the TACACS+ server and is started as part of auditd.
libtac2_1.4.0-cl3u2Basic TACACS+ server utility and communications routines.
libnss-tacplus_1.0.1-cl3u3Provides an interface between libc username lookups, the mapping functions, and the TACACS+ server.
tacplus-auth-1.0.0-cl3u1This package includes the tacplus-restrict setup utility, which enables you to perform per-command TACACS+ authorization. Per-command authorization is not done by default.
libpam-tacplus_1.4.0-1-cl3u2A modified version of the standard Debian package.
libtacplus-map1_1.0.0-cl3u2The mapping functionality between local and TACACS+ users on the server. Sets the immutable sessionid and auditing UID to ensure the original user can be tracked through multiple processes and privilege changes. Sets the auditing loginuid as immutable if supported. Creates and maintains a status database in /run/tacacs_client_map to manage and lookup mappings.
libsimple-tacacct1_1.0.0-cl3u2Provides an interface for programs to send accounting records to the TACACS+ server. Used by audisp-tacplus.
libtac2-bin_1.4.0-cl3u2Provides the tacc testing program and TACACS+ man page.

Limitations

TACACS+ Client Is only Supported through the Management Interface

The TACACS+ client is only supported through the management interface on the switch: eth0, eth1, or the VRF management interface. The TACACS+ client is not supported through bonds, switch virtual interfaces (SVIs), or switch port interfaces (swp).

Multiple TACACS+ Users

If two or more TACACS+ users are logged in simultaneously with the same privilege level, while the accounting records are maintained correctly, a lookup on either name will match both users, while a UID lookup will only return the user that logged in first.

This means that any processes run by either user will be attributed to both, and all files created by either user will be attributed to the first name matched. This is similar to adding two local users to the password file with the same UID and GID, and is an inherent limitation of using the UID for the base user from the password file.

The current algorithm returns the first name matching the UID from the mapping file; this can be the first or the second user that logged in.

To work around this issue, you can use the switch audit log or the TACACS server accounting logs to determine which processes and files are created by each user.

The Linux auditd system does not always generate audit events for processes when terminated with a signal (with the kill system call or internal errors such as SIGSEGV). As a result, processes that exit on a signal that is not caught and handled, might not generate a STOP accounting record.

Issues with deluser Command

TACACS+ and other non-local users that run the deluser command with the --remove-home option will see an error about not finding the user in /etc/passwd:

tacuser0@switch: deluser --remove-home USERNAME
userdel: cannot remove entry 'USERNAME' from /etc/passwd
/usr/sbin/deluser: `/usr/sbin/userdel USERNAME' returned error code 1. Exiting

However, the command does remove the home directory. The user can still log in on that account, but will not have a valid home directory. This is a known upstream issue with the deluser command for all non-local users.

Only use the --remove-home option when the user_homedir=1 configuration command is in use.

When Both TACACS+ and RADIUS AAA Clients are Installed

When you have both the TACACS+ and the RADIUS AAA client installed, RADIUS login is not attempted. As a workaround, do not install both the TACACS+ and the RADIUS AAA client on the same switch.

RADIUS AAA

Cumulus Networks offers add-on packages that enable RADIUS users to log in to Cumulus Linux switches in a transparent way with minimal configuration. There is no need to create accounts or directories on the switch. Authentication is handled with PAM and includes login, ssh, sudo and su.

Install the RADIUS Packages

You can install the RADIUS packages even if the switch is not connected to the internet, as they are contained in the cumulus-local-apt-archive repository that is embedded in the Cumulus Linux disk image.

To install the RADIUS packages:

cumulus@switch:~$ sudo apt-get update
cumulus@switch:~$ sudo apt-get install libnss-mapuser libpam-radius-auth

After installation is complete, either reboot the switch or run the sudo systemctl restart netd command.

The libpam-radius-auth package supplied with the Cumulus Linux RADIUS client is a newer version than the one in Debian Buster. This package contains support for IPv6, the src_ip option described below, as well as a number of bug fixes and minor features. The package also includes VRF support, provides man pages describing the PAM and RADIUS configuration, and sets the SUDO_PROMPT environment variable to the login name for RADIUS mapping support.

The libnss-mapuser package is specific to Cumulus Linux and supports the getgrent, getgrnam and getgrgid library interfaces. These interfaces add logged in RADIUS users to the group member list for groups that contain the mapped_user (radius_user) if the RADIUS account is unprivileged, and add privileged RADIUS users to the group member list for groups that contain the mapped_priv_user (radius_priv_user) during the group lookups.

During package installation:

Configure the RADIUS Client

To configure the RADIUS client, edit the /etc/pam_radius_auth.conf file:

  1. Add the hostname or IP address of at least one RADIUS server (such as a freeradius server on Linux), and the shared secret used to authenticate and encrypt communication with each server.

    The hostname of the switch must be resolvable to an IP address, which, in general, is fixed in DNS. If for some reason you cannot find the hostname in DNS, you can add the hostname to the /etc/hosts file manually. However, this can cause problems since the IP address is usually assigned by DHCP, which can change at any time.

    Multiple server configuration lines are verified in the order listed. Other than memory, there is no limit to the number of RADIUS servers you can use.

    The server port number or name is optional. The system looks up the port in the /etc/services file. However, you can override the ports in the /etc/pam_radius_auth.conf file.

  2. If the server is slow or latencies are high, change the timeout setting. The setting defaults to 3 seconds.

  3. If you want to use a specific interface to reach the RADIUS server, specify the src_ip option. You can specify the hostname of the interface, an IPv4, or an IPv6 address. If you specify the src_ip option, you must also specify the timeout option.

  4. Set the vrf-name field. This is typically set to mgmt if you are using a management VRF. You cannot specify more than one VRF.

The configuration file includes the mapped_priv_user field that sets the account used for privileged RADIUS users and the priv-lvl field that sets the minimum value for the privilege level to be considered a privileged login (the default value is 15). If you edit these fields, make sure the values match those set in the /etc/nss_mapuser.conf file.

The following example provides a sample /etc/pam_radius_auth.conf file configuration:

mapped_priv_user   radius_priv_user
# server[:port]    shared_secret   timeout (secs)  src_ip
192.168.0.254      secretkey
other-server       othersecret     3               192.168.1.10
# when mgmt vrf is in use
vrf-name mgmt

If this is the first time you are configuring the RADIUS client, uncomment the debug line to help with troubleshooting. The debugging messages are written to /var/log/syslog. When the RADIUS client is working correctly, comment out the debug line.

As an optional step, you can set PAM configuration keywords by editing the /usr/share/pam-configs/radius file. After you edit the file, you must run the pam-auth-update --package command. PAM configuration keywords are described in the pam_radius_auth (8) man page.

The privilege level for the user on the switch is determined by the value of the VSA (Vendor Specific Attribute) shell:priv-lvl. If the attribute is not returned, the user is unprivileged. The following shows an example using the freeradius server for a fully-privileged user.

Service-Type = Administrative-User,
Cisco-AVPair = "shell:roles=network-administrator",
Cisco-AVPair += "shell:priv-lvl=15"

The VSA vendor name (Cisco-AVPair in the example above) can have any content. The RADIUS client only checks for the string shell:priv-lvl.

Enable Login without Local Accounts

Because LDAP is not commonly used with switches and adding accounts locally is cumbersome, Cumulus Linux includes a mapping capability with the libnss-mapuser package.

Mapping is done using two NSS (Name Service Switch) plugins, one for account name, and one for UID lookup. These accounts are configured automatically in /etc/nsswitch.conf during installation and are removed when the package is removed. See the nss_mapuser (8) man page for the full description of this plugin.

A username is mapped at login to a fixed account specified in the configuration file, with the fields of the fixed account used as a template for the user that is logging in.

For example, if the name being looked up is dave and the fixed account in the configuration file is radius_user, and that entry in /etc/passwd is:

radius_user:x:1017:1002:radius user:/home/radius_user:/bin/bash

then the matching line returned by running getent passwd dave is:

cumulus@switch:~$ getent passwd dave
dave:x:1017:1002:dave mapped user:/home/dave:/bin/bash

The home directory /home/dave is created during the login process if it does not already exist and is populated with the standard skeleton files by the mkhomedir_helper command.

The configuration file /etc/nss_mapuser.conf is used to configure the plugins. The file includes the mapped account name, which is radius_user by default. You can change the mapped account name by editing the file. The nss_mapuser (5) man page describes the configuration file.

A flat file mapping is done based on the session number assigned during login, which persists across su and sudo. The mapping is removed at logout.

Local Fallback Authentication

If a site wants to allow local fallback authentication for a user when none of the RADIUS servers can be reached you can add a privileged user account as a local account on the switch. The local account must have the same unique identifier as the privileged user and the shell must be the same.

To configure local fallback authentication:

  1. dd a local privileged user account. For example, if the radius_priv_user account in the /etc/passwd file is radius_priv_user:x:1002:1001::/home/radius_priv_user:/sbin/radius_shell, run the following command to add a local privileged user account named johnadmin:
cumulus@switch:~$ sudo useradd -u 1002 -g 1001 -o -s /sbin/radius_shell johnadmin
  1. To enable the local privileged user to run sudo and NCLU commands, run the following commands:
cumulus@switch:~$ sudo adduser johnadmin netedit
cumulus@switch:~$ sudo adduser johnadmin sudo
cumulus@switch:~$ sudo systemctl restart netd
  1. Edit the /etc/passwd file to move the local user line before to the radius_priv_user line:
cumulus@switch:~$ sudo vi /etc/passwd
...
johnadmin:x:1002:1001::/home/johnadmin:/sbin/radius_shell
radius_priv_user:x:1002:1001::/home/radius_priv_user:/sbin/radius_shell
  1. To set the local password for the local user, run the following command:
cumulus@switch:~$ sudo passwd johnadmin

Verify RADIUS Client Configuration

To verify that the RADIUS client is configured correctly, log in as a non-privileged user and run a net add interface command.

In this example, the ops user is not a privileged RADIUS user so they cannot add an interface.

ops@leaf01:~$ net add interface swp1
ERROR: User ops does not have permission to make networking changes.

In this example, the admin user is a privileged RADIUS user (with privilege level 15) so is able to add interface swp1.

admin@leaf01:~$ net add interface swp1
admin@leaf01:~$ net pending
--- /etc/network/interfaces    2018-04-06 14:49:33.099331830 +0000
+++ /var/run/nclu/iface/interfaces.tmp    2018-04-06 16:01:16.057639999 +0000
@@ -3,10 +3,13 @@

  source /etc/network/interfaces.d/*.intf

  # The loopback network interface
  auto lo
  iface lo inet loopback

  # The primary network interface
  auto eth0
  iface eth0 inet dhcp
+
+auto swp1
iface swp1
...

Remove RADIUS Client Packages

Remove the RADIUS packages with the following command:

cumulus@switch:~$ sudo apt-get remove libnss-mapuser libpam-radius-auth

When you remove the packages, the plugins are removed from the /etc/nsswitch.conf file and from the PAM files.

To remove all configuration files for these packages, run:

cumulus@switch:~$ sudo apt-get purge libnss-mapuser libpam-radius-auth

The RADIUS fixed account is not removed from the /etc/passwd or /etc/group file and the home directories are not removed. They remain in case there are modifications to the account or files in the home directories.

To remove the home directories of the RADIUS users, first get the list by running:

cumulus@switch:~$ sudo ls -l /home | grep radius

For all users listed, except the radius_user, run this command to remove the home directories:

cumulus@switch:~$ sudo deluser --remove-home USERNAME

where USERNAME is the account name (the home directory relative portion). This command gives the following warning because the user is not listed in the /etc/passwd file.

userdel: cannot remove entry 'USERNAME' from /etc/passwd
/usr/sbin/deluser: `/usr/sbin/userdel USERNAME' returned error code 1. Exiting.

After removing all the RADIUS users, run the command to remove the fixed account. If the account has been changed in the /etc/nss_mapuser.conf file, use that account name instead of radius_user.

cumulus@switch:~$ sudo deluser --remove-home radius_user
cumulus@switch:~$ sudo deluser --remove-home radius_priv_user
cumulus@switch:~$ sudo delgroup radius_users

Limitations

Netfilter - ACLs

Netfilter is the packet filtering framework in Cumulus Linux as well as most other Linux distributions. There are a number of tools available for configuring ACLs in Cumulus Linux:

NCLU and cl-acltool operate on various configuration files and use iptables, ip6tables, and ebtables to install rules into the kernel. In addition, NCLU and cl-acltool program rules in hardware for interfaces involving switch port interfaces, which iptables, ip6tables and ebtables cannot do on their own.

In many instances, you can use NCLU to configure ACLs; however, in some cases, you must use cl-acltool. The examples below specify when to use which tool.

If you need help to configure ACLs, run net example acl to see a basic configuration:

Example
cumulus@leaf01:~$ net example acl

Scenario
========
We would like to use access-lists on 'switch' to
- Restrict inbound traffic on swp1 to traffic from 10.1.1.0/24 destined for 10.1.2.0/24
- Restrict outbound traffic on swp2 to http, https, or ssh

         *switch
            /\
      swp1 /  \ swp2
          /    \
         /      \
     host-11   host-12

switch net commands
====================

Create an ACL that accepts traffic from 10.1.1.0/24 destined for 10.1.2.0/24 and drops all other traffic

switch# net add acl ipv4 MYACL accept source-ip 10.1.1.0/24 dest-ip 10.1.2.0/24
switch# net add acl ipv4 MYACL drop source-ip any dest-ip any

Apply MYACL inbound on swp1

switch# net add interface swp1 acl ipv4 MYACL inbound

Create an ACL that accepts http, https, or ssh traffic and drops all other traffic.

switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port any dest-ip any dest-port http
switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port http dest-ip any dest-port any
switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port any dest-ip any dest-port https
switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port https dest-ip any dest-port any
switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port any dest-ip any dest-port ssh
switch# net add acl ipv4 WEB_OR_SSH accept tcp source-ip any source-port ssh dest-ip any dest-port any
switch# net add acl ipv4 WEB_OR_SSH drop source-ip any dest-ip any

Apply WEB_OR_SSH outbound on swp2
switch# net add interface swp2 acl ipv4 WEB_OR_SSH outbound

commit the staged changes
switch# net commit

Verification
============
switch# net show configuration acl

Traffic Rules In Cumulus Linux

Chains

Netfilter describes the mechanism for which packets are classified and controlled in the Linux kernel. Cumulus Linux uses the Netfilter framework to control the flow of traffic to, from, and across the switch. Netfilter does not require a separate software daemon to run; it is part of the Linux kernel itself. Netfilter asserts policies at layers 2, 3 and 4 of the OSI model by inspecting packet and frame headers based on a list of rules. Rules are defined using syntax provided by the iptables, ip6tables and ebtables userspace applications.

The rules created by these programs inspect or operate on packets at several points in the life of the packet through the system. These five points are known as chains and are shown here:

The chains and their uses are:

Tables

When building rules to affect the flow of traffic, the individual chains can be accessed by tables. Linux provides three tables by default:

Each table has a set of default chains that can be used to modify or inspect packets at different points of the path through the switch. Chains contain the individual rules to influence traffic. Each table and the default chains they support are shown below. Tables and chains in green are supported by Cumulus Linux, those in red are not supported (that is, they are not hardware accelerated) at this time.

Rules

Rules are the items that actually classify traffic to be acted upon. Rules are applied to chains, which are attached to tables, similar to the graphic below.

Rules have several different components; the examples below highlight those different components.

How Rules Are Parsed and Applied

All the rules from each chain are read from iptables, ip6tables, and ebtables and entered in order into either the filter table or the mangle table. The rules are read from the kernel in the following order:

When rules are combined and put into one table, the order determines the relative priority of the rules; iptables and ip6tables have the highest precedence and ebtables has the lowest.

The Linux packet forwarding construct is an overlay for how the silicon underneath processes packets. Be aware of the following:

-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport $BFD_ECHO_PORT -j SETCLASS --class 7
-A $INGRESS_CHAIN -p udp --dport $BFD_ECHO_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000

If multiple contiguous rules with the same match criteria are applied to --in-interface, only the first rule gets processeand then terminates processing. This is a misconfiguration; there is no reason to have duplicate rules with different actions.

-A FORWARD -i $PORTA -j ACCEPT
-A FORWARD -o $PORTA -j ACCEPT   <-- This rule is performed LAST (because of egress interface matching)
-A FORWARD -i $PORTB -j DROP
If you modify the rules like this, they are performed in order:
-A FORWARD -i $PORTA -j ACCEPT
-A FORWARD -i swp+ -o $PORTA -j ACCEPT   <-- These rules are performed in order (because of wildcard match on ingress interface)
-A FORWARD -i $PORTB -j DROP

On Broadcom switches, the ingress INPUT chain rules match layer 2 and layer 3 multicast packets before multicast packet replication has occurred; therefore, a DROP rule affects all copies.

Rule Placement in Memory

INPUT and ingress (FORWARD -i) rules occupy the same memory space. A rule counts as ingress if the -i option is set. If both input and output options (-i and -o) are set, the rule is considered as ingress and occupies that memory space. For example:

-A FORWARD -i swp1 -o swp2 -s 10.0.14.2 -d 10.0.15.8 -p tcp -j ACCEPT

If you set an output flag with the INPUT chain, you see an error. For example, running cl-acltool -i on the following rule:

-A FORWARD,INPUT -i swp1 -o swp2 -s 10.0.14.2 -d 10.0.15.8 -p tcp -j ACCEPT

generates the following error:

error: line 2 : output interface specified with INPUT chain error processing rule '-A FORWARD,INPUT -i swp1 -o swp2 -s 10.0.14.2 -d 10.0.15.8 -p tcp -j ACCEPT'

However, removing the -o option and interface make it a valid rule.

Nonatomic Update Mode and Atomic Update Mode

In Cumulus Linux, atomic update mode is enabled by default. However, this mode limits the number of ACL rules that you can configure.

To increase the number of ACL rules that can be configured, configure the switch to operate in nonatomic mode.

How the Rules Get Installed

Instead of reserving 50% of your TCAM space for atomic updates, incremental update uses the available free space to write the new TCAM rules and swap over to the new rules after this is complete. Cumulus Linux then deletes the old rules and frees up the original TCAM space. If there is insufficient free space to complete this task, the original nonatomic update is performed, which interrupts traffic.

Enable Nonatomic Update Mode

You can enable nonatomic updates for switchd, which offer better scaling because all TCAM resources are used to actively impact traffic. With atomic updates, half of the hardware resources are on standby and do not actively impact traffic.

Incremental nonatomic updates are table based, so they do not interrupt network traffic when new rules are installed. The rules are mapped into the following tables and are updated in this order:

Only Broadcom-based ASICs support incremental nonatomic updates. Mellanox Spectrum-based ASICs do not support incremental updates; therefore traffic is affected and gets stopped.

The incremental nonatomic update operation follows this order:

  1. Updates are performed incrementally, one table at a time without stopping traffic.
  2. Cumulus Linux checks if the rules in a table have changed since the last time they were installed; if a table does not have any changes, it is not reinstalled.
  3. If there are changes in a table, the new rules are populated in new groups or slices in hardware, then that table is switched over to the new groups or slices.
  4. Finally, old resources for that table are freed. This process is repeated for each of the tables listed above.
  5. If sufficient resources do not exist to hold both the new rule set and old rule set, the regular nonatomic mode is attempted. This interrupts network traffic.
  6. If the regular nonatomic update fails, Cumulus Linux reverts back to the previous rules.

To always start switchd with nonatomic updates:

  1. Edit /etc/cumulus/switchd.conf.
  2. Add the following line to the file:
acl.non_atomic_update_mode = TRUE
  1. Restart switchd:
cumulus@switch:~$ sudo systemctl restart switchd.service

During regular non-incremental nonatomic updates, traffic is stopped first, then enabled after the new configuration is written into the hardware completely.

Use iptables, ip6tables, and ebtables Directly

Using iptables, ip6tables, ebtables directly is not recommended because any rules installed in these cases only are applied to the Linux kernel and are not hardware accelerated using synchronization to the switch silicon. Running cl-acltool -i (the installation command) resets all rules and deletes anything that is not stored in /etc/cumulus/acl/policy.conf.

For example, performing:

cumulus@switch:~$ sudo iptables -A INPUT -p icmp --icmp-type echo-request -j DROP

Appears to work, and the rule appears when you run cl-acltool -L:

cumulus@switch:~$ sudo cl-acltool -L ip
-------------------------------
Listing rules of type iptables:
-------------------------------

TABLE filter :
Chain INPUT (policy ACCEPT 72 packets, 5236 bytes)
pkts bytes target prot opt in out source destination
0 0 DROP icmp -- any any anywhere anywhere icmp echo-request

However, the rule is not synced to hardware when applied in this way and running cl-acltool -i or reboot removes the rule without replacing it. To ensure all rules that can be in hardware are hardware accelerated, place them in /etc/cumulus/acl/policy.conf and install them by running cl-acltool -i.

Estimate the Number of Rules

To estimate the number of rules you can create from an ACL entry, first determine if that entry is an ingress or an egress. Then, determine if it is an IPv4-mac or IPv6 type rule. This determines the slice to which the rule belongs. Use the following to determine how many entries are used up for each type.

By default, each entry occupies one double wide entry, except if the entry is one of the following:

-A FORWARD --in-interface swp+ --out-interface swp1s0,swp1s1 -p icmp -j ACCEPT
-A FORWARD --in-interface swp1s0,swp1s1 --out-interface swp1s2,swp1s3 -p icmp -j ACCEPT
-A FORWARD --in-interface swp+ -p tcp -m multiport --dports 1050:1051,1055:1056 -j ACCEPT

Port ranges are only allowed for ingress rules.

Match SVI and Bridged Interfaces in Rules

Cumulus Linux supports matching ACL rules for both ingress and egress interfaces on both VLAN-aware and traditional mode bridges, including bridge SVIs (switch VLAN interfaces) for input and output. However, keep the following in mind:

Example rules for a VLAN-aware bridge:

[ebtables]
-A FORWARD -i br0.100 -p IPv4 --ip-protocol icmp -j DROP
-A FORWARD -o br0.100 -p IPv4 --ip-protocol icmp -j ACCEPT

iptables]
-A FORWARD -i br0.100 -p icmp -j DROP
-A FORWARD --out-interface br0.100 -p icmp -j ACCEPT
-A FORWARD --in-interface br0.100 -j POLICE --set-mode  pkt  --set-rate  1 --set-burst 1 --set-class 0

Example rules for a traditional mode bridge:

[ebtables]
-A FORWARD -i br0 -p IPv4 --ip-protocol icmp -j DROP
-A FORWARD -o br0 -p IPv4 --ip-protocol icmp -j ACCEPT

[iptables]
-A FORWARD -i br0 -p icmp -j DROP
-A FORWARD --out-interface br0 -p icmp -j ACCEPT
-A FORWARD --in-interface br0 -j POLICE --set-mode  pkt  --set-rate  1 --set-burst 1 --set-class 0

Match on VLAN IDs on Layer 2 Interfaces

On switches with Spectrum ASICs, you can match on VLAN IDs on layer 2 interfaces for ingress rules.

The following example matches on a VLAN and DSCP class, and sets the internal class of the packet. This can be combined with ingress iptable rules to get extended matching on IP fields.

[ebtables]
-A FORWARD -p 802_1Q --vlan-id 100 -j mark --mark-set 0x66

[iptables]
-A FORWARD -i swp31 -m mark --mark 0x66 -m dscp --dscp-class CS1 -j SETCLASS --class 2

Install and Manage ACL Rules with NCLU

NCLU provides an easy way to create custom ACLs in Cumulus Linux. The rules you create live in the /var/lib/cumulus/nclu/nclu_acl.conf file, which gets converted to a rules file, /etc/cumulus/acl/policy.d/50_nclu_acl.rules. This way, the rules you create with NCLU are independent of the two default files in /etc/cumulus/acl/policy.d/ 00control_plane.rules and 99control_plane_catch_all.rules, as the content in these files might get updated after you upgrade Cumulus Linux.

Instead of crafting a rule by hand then installing it using cl-acltool, NCLU handles many of the options automatically. For example, consider the following iptables rule:

-A FORWARD -i swp1 -o swp2 -s 10.0.14.2 -d 10.0.15.8 -p tcp -j ACCEPT

You create this rule, called EXAMPLE1, using NCLU like this:

cumulus@switch:~$ net add acl ipv4 EXAMPLE1 accept tcp source-ip 10.0.14.2/32 source-port any dest-ip 10.0.15.8/32 dest-port any
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

All options, such as the -j and -p, even FORWARD in the above rule, are added automatically when you apply the rule to the control plane; NCLU figures it all out for you.

You can also set a priority value, which specifies the order in which the rules get executed and the order in which they appear in the rules file. Lower numbers are executed first. To add a new rule in the middle, first run net show config acl, which displays the priority numbers. Otherwise, new rules get appended to the end of the list of rules in the nclu_acl.conf and 50_nclu_acl.rules files.

If you need to hand edit a rule, do not edit the 50_nclu_acl.rules file. Instead, edit the nclu_acl.conf file.

After you add the rule, you need to apply it to an inbound or outbound interface using net add int acl. The inbound interface in our example is swp1:

cumulus@switch:~$ net add int swp1 acl ipv4 EXAMPLE1 inbound
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

After you commit your changes, you can verify the rule you created with NCLU by running net show configuration acl:

cumulus@switch:~$ net show configuration acl
acl ipv4 EXAMPLEv4 priority 10 accept tcp source-ip 10.0.14.2/32 source-port any dest-ip 10.0.15.8/32 dest-port any

interface swp1
acl ipv4 EXAMPLE1 inbound

Or you can see all of the rules installed by running cat on the 50_nclu_acl.rules file:

cumulus@switch:~$ cat /etc/cumulus/acl/policy.d/50_nclu_acl.rules
[iptables]
# swp1: acl ipv4 EXAMPLE1 inbound
-A FORWARD --in-interface swp1 --out-interface swp2 -j ACCEPT -p tcp -s 10.0.14.2/32 -d 10.0.15.8/32 --dport 110

For INPUT and FORWARD rules, apply the rule to a control plane interface using net add control-plane:

cumulus@switch:~$ net add control-plane acl ipv4 EXAMPLE1 inbound
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To remove a rule, use net del acl ipv4|ipv6|mac RULENAME:

cumulus@switch:~$ net del acl ipv4 EXAMPLE1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

This deletes all rules from the 50_nclu_acl.rules file with that name. It also deletes the interfaces referenced in the nclu_acl.conf file.

Install and Manage ACL Rules with cl-acltool

You can manage Cumulus Linux ACLs with cl-acltool. Rules are first written to the iptables chains, as described above, and then synced to hardware via switchd.

Use iptables/ip6tables/ebtables and cl-acltool to manage rules in the default files, 00control_plane.rules and 99control_plane_catch_all.rules; they are not aware of rules created using NCLU.

To examine the current state of chains and list all installed rules, run:

cumulus@switch:~$ sudo cl-acltool -L all
 -------------------------------
Listing rules of type iptables:
-------------------------------

TABLE filter :
Chain INPUT (policy ACCEPT 90 packets, 14456 bytes)
pkts bytes target prot opt in out source destination
0 0 DROP all -- swp+ any 240.0.0.0/5 anywhere
0 0 DROP all -- swp+ any loopback/8 anywhere
0 0 DROP all -- swp+ any base-address.mcast.net/8 anywhere
0 0 DROP all -- swp+ any 255.255.255.255 anywhere ...

To list installed rules using native iptables, ip6tables and ebtables, use the -L option with the respective commands:

cumulus@switch:~$ sudo iptables -L
cumulus@switch:~$ sudo ip6tables -L
cumulus@switch:~$ sudo ebtables -L

To flush all installed rules, run:

cumulus@switch:~$ sudo cl-acltool -F all

To flush only the IPv4 iptables rules, run:

cumulus@switch:~$ sudo cl-acltool -F ip

If the install fails, ACL rules in the kernel and hardware are rolled back to the previous state. Errors from programming rules in the kernel or ASIC are reported appropriately.

Install Packet Filtering (ACL) Rules

cl-acltool takes access control list (ACL) rules input in files. Each ACL policy file contains iptables, ip6tables and ebtables categories under the tags [iptables], [ip6tables] and [ebtables].

Each rule in an ACL policy must be assigned to one of the rule categories above.

See man cl-acltool(5) for ACL rule details. For iptables rule syntax, see man iptables(8). For ip6tables rule syntax, see man ip6tables(8). For ebtables rule syntax, see man ebtables(8).

See man cl-acltool(5) and man cl-acltool(8) for further details on using cl-acltool. Some examples are listed here and more are listed later in this chapter.

By default:

  • ACL policy files are located in /etc/cumulus/acl/policy.d/.
  • All *.rules files in this directory are included in /etc/cumulus/acl/policy.conf.
  • All files included in this policy.conf file are installed when the switch boots up.
  • The policy.conf file expects rules files to have a .rules suffix as part of the file name.

Here is an example ACL policy file:

[iptables]
-A INPUT --in-interface swp1 -p tcp --dport 80 -j ACCEPT
-A FORWARD --in-interface swp1 -p tcp --dport 80 -j ACCEPT

[ip6tables]
-A INPUT --in-interface swp1 -p tcp --dport 80 -j ACCEPT
-A FORWARD --in-interface swp1 -p tcp --dport 80 -j ACCEPT

[ebtables]
-A INPUT -p IPv4 -j ACCEPT
-A FORWARD -p IPv4 -j ACCEPT

You can use wildcards or variables to specify chain and interface lists to ease administration of rules.

Currently only swp+ and bond+ are supported as wildcard names. There might be kernel restrictions in supporting more complex wildcards like swp1+ etc.

swp+ rules are applied as an aggregate, not per port. If you want to apply per port policing, specify a specific port instead of the wildcard.

INGRESS = swp+
INPUT_PORT_CHAIN = INPUT,FORWARD

[iptables]
-A $INPUT_PORT_CHAIN --in-interface $INGRESS -p tcp --dport 80 -j ACCEPT

[ip6tables]
-A $INPUT_PORT_CHAIN --in-interface $INGRESS -p tcp --dport 80 -j ACCEPT

[ebtables]
-A INPUT -p IPv4 -j ACCEPT

You can write ACL rules for the system into multiple files under the default /etc/cumulus/acl/policy.d/ directory. The ordering of rules during installation follows the sort order of the files based on their file names.

Use multiple files to stack rules. The example below shows two rules files separating rules for management and datapath traffic:

cumulus@switch:~$ ls /etc/cumulus/acl/policy.d/
00sample_mgmt.rules 01sample_datapath.rules
cumulus@switch:~$ cat /etc/cumulus/acl/policy.d/00sample_mgmt.rules

INGRESS_INTF = swp+
INGRESS_CHAIN = INPUT

[iptables]
# protect the switch management
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -s 10.0.14.2 -d 10.0.15.8 -p tcp -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -s 10.0.11.2 -d 10.0.12.8 -p tcp -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -d 10.0.16.8 -p udp -j DROP

cumulus@switch:~$ cat /etc/cumulus/acl/policy.d/01sample_datapath.rules
INGRESS_INTF = swp+
INGRESS_CHAIN = INPUT, FORWARD

[iptables]
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -s 192.0.2.5 -p icmp -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -s 192.0.2.6 -d 192.0.2.4 -j DROP
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -s 192.0.2.2 -d 192.0.2.8 -j DROP

Install all ACL policies under a directory:

cumulus@switch:~$ sudo cl-acltool -i -P ./rules
Reading files under rules
Reading rule file ./rules/01_http_rules.txt ...
Processing rules in file ./rules/01_http_rules.txt ...
Installing acl policy ...
Done.

Install all rules and policies included in /etc/cumulus/acl/policy.conf:

cumulus@switch:~$ sudo cl-acltool -i

Specify the Policy Files to Install

By default, Cumulus Linux installs any .rules file you configure in /etc/cumulus/acl/policy.d/. To add other policy files to an ACL, you need to include them in /etc/cumulus/acl/policy.conf. For example, for Cumulus Linux to install a rule in a policy file called 01_new.datapathacl, add include /etc/cumulus/acl/policy.d/01_new.rules to policy.conf, as in this example:

cumulus@switch:~$ sudo nano /etc/cumulus/acl/policy.conf

#
# This file is a master file for acl policy file inclusion
#
# Note: This is not a file where you list acl rules.
#
# This file can contain:
# - include lines with acl policy files
#   example:
#     include <filepath>
#
# see manpage cl-acltool(5) and cl-acltool(8) for how to write policy files 
#

include /etc/cumulus/acl/policy.d/01_new.datapathacl

Hardware Limitations on Number of Rules

The maximum number of rules that can be handled in hardware is a function of the following factors:

If the maximum number of rules for a particular table is exceeded, cl-acltool -i generates the following error:

error: hw sync failed (sync_acl hardware installation failed) Rolling back .. failed.

In the tables below, the default rules count toward the limits listed. The raw limits below assume only one ingress and one egress table are present.

Broadcom Tomahawk Limits

DirectionAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
Ingress raw limit51251210241024
Ingress limit with default rules256 (36 default)256 (29 default)768 (36 default)768 (29 default)
Egress raw limit25605120
Egress limit with default rules256 (29 default)0512 (29 default)0

Broadcom Trident3 Limits

The Trident3 ASIC is divided into 12 slices, organized into 4 groups for ACLs. Each group contains 3 slices. Each group can support a maximum of 768 rules. You cannot mix IPv4 and IPv6 rules within the same group. IPv4 and MAC rules can be programmed into the same group.

DirectionAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
Ingress raw limit76876823042304
Ingress limit with default rules768 (44 default)768 (41 default)2304 (44 default)2304 (41 default)
Egress raw limit51205120
Egress limit with default rules512 (28 default)0512 (28 default)0

Broadcom Trident II+ Limits

DirectionAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
Ingress raw limit4096409681928192
Ingress limit with default rules2048 (36 default)3072 (29 default)6144 (36 default)6144 (29 default)
Egress raw limit25605120
Egress limit with default rules256 (29 default)0512 (29 default)0

Broadcom Trident II Limits

DirectionAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
Ingress raw limit1024102420482048
Ingress limit with default rules512 (36 default)768 (29 default)1536 (36 default)1536 (29 default)
Egress raw limit25605120
Egress limit with default rules256 (29 default)0512 (29 default)0

Broadcom Helix4 Limits

DirectionAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
Ingress raw limit102451220481024
Ingress limit with default rules768 (36 default)384 (29 default)1792 (36 default)896 (29 default)
Egress raw limit25605120
Egress limit with default rules256 (29 default)0512 (29 default)0

Mellanox Spectrum Limits

The Mellanox Spectrum ASIC has one common TCAM for both ingress and egress, which can be used for other non-ACL-related resources. However, the number of supported rules varies with the TCAM profile specified for the switch.

ProfileAtomic Mode IPv4 RulesAtomic Mode IPv6 RulesNonatomic Mode IPv4 RulesNonatomic Mode IPv6 Rules
default5002501000500
ipmc-heavy75050015001000
acl-heavy1750100035002000
ipmc-max100050020001000
ip-acl-heavy75000150000

Even though the table above specifies that zero IPv6 rules are supported with the ip-acl-heavy profile, Cumulus Networks does not prevent you from configuring IPv6 rules. However, there is no guarantee that IPv6 rules work under the ip-acl-heavy profile.

Supported Rule Types

The iptables/ip6tables/ebtables construct tries to layer the Linux implementation on top of the underlying hardware but they are not always directly compatible. Here are the supported rules for chains in iptables, ip6tables and ebtables.

To learn more about any of the options shown in the tables below, run iptables -h [name of option]. The same help syntax works for options for ip6tables and ebtables.

Click to see an example of help syntax for an ebtables target
root@leaf1# ebtables -h tricolorpolice
<...snip...>
tricolorpolice option:
--set-color-mode STRING setting the mode in blind or aware
--set-cir INT setting committed information rate in kbits per second
--set-cbs INT setting committed burst size in kbyte
--set-pir INT setting peak information rate in kbits per second
--set-ebs INT setting excess burst size in kbyte
--set-conform-action-dscp INT setting dscp value if the action is accept for conforming packets
--set-exceed-action-dscp INT setting dscp value if the action is accept for exceeding packets
--set-violate-action STRING setting the action (accept/drop) for violating packets
--set-violate-action-dscp INT setting dscp value if the action is accept for violating packets
Supported chains for the filter table:
INPUT FORWARD OUTPUT

iptables and ip6tables Rule Support

Rule ElementSupportedUnsupported
MatchesSrc/Dst, IP protocol
In/out interface
IPv4: icmp, ttl,
IPv6: icmp6, frag, hl,
IP common: tcp (with flags), udp, multiport, DSCP, addrtype
Rules with input/output Ethernet interfaces are ignored
Inverse matches
Standard TargetsACCEPT, DROPRETURN, QUEUE, STOP, Fall Thru, Jump
Extended TargetsLOG (IPv4/IPv6); UID is not supported for LOG
TCP SEQ, TCP options or IP options
ULOG
SETQOS
DSCP
Unique to Cumulus Linux:
SPAN
ERSPAN (IPv4/IPv6)
POLICE
TRICOLORPOLICE
SETCLASS

ebtables Rule Support

Rule ElementSupportedUnsupported
Matchesether type
input interface/wildcard
output interface/wildcard
src/dst MAC
IP: src, dest, tos, proto, sport, dport
IPv6: tclass, icmp6: type, icmp6: code range, src/dst addr, sport, dport
802.1p (CoS)
VLAN
Inverse matches
Proto length
Standard TargetsACCEPT, DROPReturn, Continue, Jump, Fall Thru
Extended TargetsUlog
log
Unique to Cumulus Linux:
span
erspan
police
tricolorpolice
setclass

Other Unsupported Rules

IPv6 Egress Rules on Broadcom Switches

Cumulus Linux supports IPv6 egress rules in ip6tables on Broadcom switches. Because there are no slices to allocate in the egress TCAM for IPv6, the matches are implemented using a combination of the ingress IPv6 slice and the existing egress IPv4 MAC slice:

For example, the -A FORWARD --out-interface vlan100 -p icmp6 -j ACCEPT rule is split into the following:

  • IPv6 egress rules in ip6tables are not supported on Hurricane2 switches.
  • You cannot match both input and output interfaces in the same rule.
  • The egress TCAM IPv4 MAC slice is shared with other rules, which constrains the scale to a much lower limit.

Caveats

Splitting rules across the ingress TCAM and the egress TCAM causes the ingress IPv6 part of the rule to match packets going to all destinations, which can interfere with the regular expected linear rule match in a sequence. For example:

A higher rule can prevent a lower rule from being matched unexpectedly:

Rule 1: -A FORWARD --out-interface vlan100 -p icmp6 -j ACCEPT

Rule 2: -A FORWARD --out-interface vlan101 -p icmp6 -s 01::02 -j ACCEPT

Rule 1 matches all icmp6 packets from to all out interfaces in the ingress TCAM.`

This prevents rule 2 from getting matched, which is more specific but with a different out interface. Make sure to put more specific matches above more general matches even if the output interfaces are different.

When you have two rules with the same output interface, the lower rule might match unexpectedly depending on the presence of the previous rules.

Rule 1: -A FORWARD --out-interface vlan100 -p icmp6 -j ACCEPT

Rule 2: -A FORWARD --out-interface vlan101 -s 00::01 -j DROP

Rule 3: -A FORWARD --out -interface vlan101 -p icmp6 -j ACCEPT

Rule 3 still matches for an icmp6 packet with sip 00:01 going out of vlan101. Rule 1 interferes with the normal function of rule 2 and/or rule 3.

When you have two adjacent rules with the same match and different output interfaces, such as:

Rule 1: -A FORWARD --out-interface vlan100 -p icmp6 -j ACCEPT

Rule 2: -A FORWARD --out-interface vlan101 -p icmp6 -j DROP

Rule 2 will never be match on ingress. Both rules share the same mark.

Matching Untagged Packets (Trident3 Switches)

Untagged packets do not have an associated VLAN to match on egress; therefore, the match must be on the underlying layer 2 port. For example, for a bridge configured with pvid 100, member port swp1s0 and swp1s1, and SVI vlan100, the output interface match on vlan100 has to be expanded into each member port. The -A FORWARD -o vlan100 -p icmp6 -j ACCEPT rule must be specified as two rules:

Rule 1: -A FORWARD -o swp1s0 -p icmp6 -J ACCEPT

Rule 2: -A FORWARD -o swp1s1 -p icmp6 -j ACCEPT

Matching on an egress port matches all packets egressing the port, tagged as well as untagged. Therefore, to match only untagged traffic on the port, you must specify additional rules above this rule to prevent tagged packets matching the rule. This is true for bridge member ports as well as regular layer 2 ports. In the example rule above, if vlan101 is also present on the bridge, add a rule above rule 1 and rule 2 to protect vlan101 tagged traffic:

Rule 0: -A FORWARD -o vlan101 -p icmp6 -j ACCEPT

Rule 1: -A FORWARD -o swp1s0 -p icmp6 -j ACCEPT

Rule 2: -A FORWARD -o swp1s1 -p icmp6 -j ACCEPT

For a standalone port or subinterface on swp1s2:

Rule 0: -A FORWARD -o swp1s2.101 -p icmp6 -j ACCEPT

Rule 1: -A FORWARD -o swp1s2 -p icmp6 -j ACCEPT

Common Examples

Control Plane and Data Plane Traffic

You can configure quality of service for traffic on both the control plane and the data plane. By using QoS policers, you can rate limit traffic so incoming packets get dropped if they exceed specified thresholds.

Counters on POLICE ACL rules in iptables do not currently show the packets that are dropped due to those rules.

Use the POLICE target with iptables. POLICE takes these arguments:

For example, to rate limit the incoming traffic on swp1 to 400 packets per second with a burst of 100 packets per second and set the class of the queue for the policed traffic as 0, set this rule in your appropriate .rules file:

-A INPUT --in-interface swp1 -j POLICE --set-mode pkt --set-rate 400 --set-burst 100 --set-class 0

Here is another example of control plane ACL rules to lock down the switch. You specify them in /etc/cumulus/acl/policy.d/00control_plane.rules:

View the contents of the file ...
INGRESS_INTF = swp+
INGRESS_CHAIN = INPUT
INNFWD_CHAIN = INPUT,FORWARD
MARTIAN_SOURCES_4 = "240.0.0.0/5,127.0.0.0/8,224.0.0.0/8,255.255.255.255/32"
MARTIAN_SOURCES_6 = "ff00::/8,::/128,::ffff:0.0.0.0/96,::1/128"

# Custom Policy Section
SSH_SOURCES_4 = "192.168.0.0/24"
NTP_SERVERS_4 = "192.168.0.1/32,192.168.0.4/32"
DNS_SERVERS_4 = "192.168.0.1/32,192.168.0.4/32"
SNMP_SERVERS_4 = "192.168.0.1/32"

[iptables]
-A $INNFWD_CHAIN --in-interface $INGRESS_INTF -s $MARTIAN_SOURCES_4 -j DROP
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p ospf -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p tcp --dport bgp -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p tcp --sport bgp -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p icmp -j POLICE --set-mode pkt --set-rate 100 --set-burst 40 --set-class 2
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport bootps:bootpc -j POLICE --set-mode pkt --set-rate 100 --set-burst 100 --set-class 2
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p tcp --dport bootps:bootpc -j POLICE --set-mode pkt --set-rate 100 --set-burst 100 --set-class 2
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p igmp -j POLICE --set-mode pkt --set-rate 300 --set-burst 100 --set-class 6

# Custom policy
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p tcp --dport 22 -s $SSH_SOURCES_4 -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --sport 123 -s $NTP_SERVERS_4 -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --sport 53 -s $DNS_SERVERS_4 -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport 161 -s $SNMP_SERVERS_4 -j ACCEPT


# Allow UDP traceroute when we are the current TTL expired hop 
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport 1024:65535 -m ttl --ttl-eq 1 -j ACCEPT
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -j DROP

Set DSCP on Transit Traffic

The examples here use the mangle table to modify the packet as it transits the switch. DSCP is expressed in decimal notation in the examples below.

[iptables]

#Set SSH as high priority traffic.
-t mangle -A FORWARD -p tcp --dport 22  -j DSCP --set-dscp 46

#Set everything coming in SWP1 as AF13
-t mangle -A FORWARD --in-interface swp1 -j DSCP --set-dscp 14

#Set Packets destined for 10.0.100.27 as best effort
-t mangle -A FORWARD -d 10.0.100.27/32 -j DSCP --set-dscp 0

#Example using a range of ports for TCP traffic
-t mangle -A FORWARD -p tcp -s 10.0.0.17/32 --sport 10000:20000 -d 10.0.100.27/32 --dport 10000:20000 -j DSCP --set-dscp 34

Verify DSCP Values on Transit Traffic

The examples here use the DSCP match criteria in combination with other IP, TCP, and interface matches to identify traffic and count the number of packets.

[iptables]

#Match and count the packets that match SSH traffic with DSCP EF
-A FORWARD -p tcp --dport 22 -m dscp --dscp 46 -j ACCEPT

#Match and count the packets coming in SWP1 as AF13
-A FORWARD --in-interface swp1 -m dscp --dscp 14 -j ACCEPT
#Match and count the packets with a destination 10.0.0.17 marked best effort
-A FORWARD -d 10.0.100.27/32 -m dscp --dscp 0 -j ACCEPT

#Match and count the packets in a port range with DSCP AF41
-A FORWARD -p tcp -s 10.0.0.17/32 --sport 10000:20000 -d 10.0.100.27/32 --dport 10000:20000 -m dscp --dscp 34 -j ACCEPT

Check the Packet and Byte Counters for ACL Rules

To verify the counters using the above example rules, first send test traffic matching the patterns through the network. The following example generates traffic with mz (or mausezahn), which can be installed on host servers or even on Cumulus Linux switches. After traffic is sent to validate the counters, they are matched on switch1 using cl-acltool.

Policing counters do not increment on switches with the Spectrum ASIC.

# Send 100 TCP packets on host1 with a DSCP value of EF with a destination of host2 TCP port 22:

cumulus@host1$ mz eth1 -A 10.0.0.17 -B 10.0.100.27 -c 100 -v -t tcp "dp=22,dscp=46"
  IP:  ver=4, len=40, tos=184, id=0, frag=0, ttl=255, proto=6, sum=0, SA=10.0.0.17, DA=10.0.100.27,
      payload=[see next layer]
  TCP: sp=0, dp=22, S=42, A=42, flags=0, win=10000, len=20, sum=0,
      payload=

# Verify the 100 packets are matched on switch1

cumulus@switch1$ sudo cl-acltool -L ip
-------------------------------
Listing rules of type iptables:
-------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 9314 packets, 753K bytes)
  pkts bytes target     prot opt in     out     source               destination
Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
  pkts bytes target     prot opt in     out     source               destination
  100  6400 ACCEPT     tcp  --  any    any     anywhere             anywhere             tcp dpt:ssh DSCP match 0x2e
    0     0 ACCEPT     all  --  swp1   any     anywhere             anywhere             DSCP match 0x0e
    0     0 ACCEPT     all  --  any    any     10.0.0.17            anywhere             DSCP match 0x00
    0     0 ACCEPT     tcp  --  any    any     10.0.0.17            10.0.100.27          tcp spts:webmin:20000
    dpts:webmin:2002

# Send 100 packets with a small payload on host1 with a DSCP value of AF13 with a destination of host2:

cumulus@host1$ mz eth1 -A 10.0.0.17 -B 10.0.100.27 -c 100 -v -t ip
  IP:  ver=4, len=20, tos=0, id=0, frag=0, ttl=255, proto=0, sum=0, SA=10.0.0.17, DA=10.0.100.27,
      payload=

# Verify the 100 packets are matched on switch1

cumulus@switch1$ sudo cl-acltool -L ip
-------------------------------
Listing rules of type iptables:
-------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 9314 packets, 753K bytes)
  pkts bytes target     prot opt in     out     source               destination
  chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
  pkts bytes target     prot opt in     out     source               destination
  100  6400 ACCEPT     tcp  --  any    any     anywhere             anywhere             tcp dpt:ssh DSCP match 0x2e
  100  7000 ACCEPT     all  --  swp3   any     anywhere             anywhere             DSCP match 0x0e
  100  6400 ACCEPT     all  --  any    any     10.0.0.17            anywhere             DSCP match 0x00
    0     0 ACCEPT     tcp  --  any    any     10.0.0.17            10.0.100.27          tcp spts:webmin:20000 dpts:webmin:2002

# Send 100 packets on host1 with a destination of host2:

cumulus@host1$ mz eth1 -A 10.0.0.17 -B 10.0.100.27 -c 100 -v -t ip
 IP:  ver=4, len=20, tos=56, id=0, frag=0, ttl=255, proto=0, sum=0, SA=10.0.0.17, DA=10.0.100.27,
     payload=

# Verify the 100 packets are matched on switch1

cumulus@switch1$ sudo cl-acltool -L ip
-------------------------------
Listing rules of type iptables:
-------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 9314 packets, 753K bytes)
  pkts bytes target     prot opt in     out     source               destination
Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
  pkts bytes target     prot opt in     out     source               destination
  100  6400 ACCEPT     tcp  --  any    any     anywhere             anywhere             tcp dpt:ssh DSCP match 0x2e
  100  7000 ACCEPT     all  --  swp3   any     anywhere             anywhere             DSCP match 0x0e
    0     0 ACCEPT     all  --  any    any     10.0.0.17            anywhere             DSCP match 0x00
    0     0 ACCEPT     tcp  --  any    any     10.0.0.17            10.0.100.27          tcp spts:webmin:20000 dpts:webmin:2002Still working

Filter Specific TCP Flags

The example solution below creates rules on the INPUT and FORWARD chains to drop ingress IPv4 and IPv6 TCP packets when the SYN bit is set and the RST, ACK, and FIN bits are reset. The default for the INPUT and FORWARD chains allows all other packets. The ACL is applied to ports swp20 and swp21. After configuring this ACL, new TCP sessions that originate from ingress ports swp20 and swp21 are not allowed. TCP sessions that originate from any other port are allowed.

INGRESS_INTF = swp20,swp21

[iptables]
-A INPUT,FORWARD --in-interface $INGRESS_INTF -p tcp --syn -j DROP
[ip6tables]
-A INPUT,FORWARD --in-interface $INGRESS_INTF -p tcp --syn -j DROP

The --syn flag in the above rule matches packets with the SYN bit set and the ACK, RST, and FIN bits are cleared. It is equivalent to using -tcp-flags SYN,RST,ACK,FIN SYN. For example, you can write the above rule as:

-A INPUT,FORWARD --in-interface $INGRESS_INTF -p tcp --tcp-flags SYN,RST,ACK,FIN SYN -j DROP

Control Who Can SSH into the Switch

Run the following NCLU commands to control who can SSH into the switch. In the following example, 10.0.0.11/32 is the interface IP address (or loopback IP address) of the switch and 10.255.4.0/24 can SSH into the switch.

cumulus@switch:~$ net add acl ipv4 test priority 10 accept source-ip 10.255.4.0/24 dest-ip 10.0.0.11/32
cumulus@switch:~$ net add acl ipv4 test priority 20 drop source-ip any dest-ip 10.0.0.11/32
cumulus@switch:~$ net add control-plane acl ipv4 test inbound
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Cumulus Linux does not support the keyword iprouter (typically used for traffic sent to the CPU, where the destination MAC address is that of the router but the destination IP address is not the router).

Example Scenario

The following example scenario demonstrates how several different rules are applied.

Following are the configurations for the two switches used in these examples. The configuration for each switch appears in /etc/network/interfaces on that switch.

Switch 1 Configuration

cumulus@switch1:~$ net show configuration files
...
/etc/network/interfaces
=======================

auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4

auto bond2
iface bond2
  bond-slaves swp3 swp4

auto br-untagged
iface br-untagged
  address 10.0.0.1/24
  bridge_ports swp1 bond2
  bridge_stp on

auto br-tag100
iface br-tag100
  address 10.0.100.1/24
  bridge_ports swp2.100 bond2.100
  bridge_stp on
...

Switch 2 Configuration

cumulus@switch2:~$ net show configuration files
...
/etc/network/interfaces
=======================

auto swp3
iface swp3

auto swp4
iface swp4

auto br-untagged
iface br-untagged
  address 10.0.0.2/24
  bridge_ports bond2
  bridge_stp on

auto br-tag100
iface br-tag100
  address 10.0.100.2/24
  bridge_ports bond2.100
  bridge_stp on

auto bond2
iface bond2
  bond-slaves swp3 swp4
...

Egress Rule

The following rule blocks any TCP traffic with destination port 200 going from host1 or host2 through the switch (corresponding to rule 1 in the diagram above).

[iptables] -A FORWARD -o bond2 -p tcp --dport 200 -j DROP

Ingress Rule

The following rule blocks any UDP traffic with source port 200 going from host1 through the switch (corresponding to rule 2 in the diagram above).

[iptables] -A FORWARD -i swp2 -p udp --sport 200 -j DROP

Input Rule

The following rule blocks any UDP traffic with source port 200 and destination port 50 going from host1 to the switch (corresponding to rule 3 in the diagram above).

[iptables] -A INPUT -i swp1 -p udp --sport 200 --dport 50 -j DROP

Output Rule

The following rule blocks any TCP traffic with source port 123 and destination port 123 going from Switch 1 to host2 (corresponding to rule 4 in the diagram above).

[iptables] -A OUTPUT -o br-tag100 -p tcp --sport 123 --dport 123 -j DROP

Combined Rules

The following rule blocks any TCP traffic with source port 123 and destination port 123 going from any switch port egress or generated from Switch 1 to host1 or host2 (corresponding to rules 1 and 4 in the diagram above).

[iptables] -A OUTPUT,FORWARD -o swp+ -p tcp --sport 123 --dport 123 -j DROP

This also becomes two ACLs and is the same as:

[iptables]
-A FORWARD -o swp+ -p tcp --sport 123 --dport 123 -j DROP 
-A OUTPUT -o swp+ -p tcp --sport 123 --dport 123 -j DROP

Layer 2-only Rules/ebtables

The following rule blocks any traffic with source MAC address 00:00:00:00:00:12 and destination MAC address 08:9e:01:ce:e2:04 going from any switch port egress/ingress.

[ebtables] -A FORWARD -s 00:00:00:00:00:12 -d 08:9e:01:ce:e2:04 -j DROP

Caveats and Errata

Not All Rules Supported

Not all iptables, ip6tables, or ebtables rules are supported. Refer to the Supported Rules section above for specific rule support.

ACL Log Policer Limits Traffic

To protect the CPU from overloading, traffic copied to the CPU is limited to 1 pkt/s by an ACL Log Policer.

Bridge Traffic Limitations

Bridge traffic that matches LOG ACTION rules are not logged in syslog; the kernel and hardware identify packets using different information.

Log Actions Cannot Be Forwarded

Logged packets cannot be forwarded. The hardware cannot both forward a packet and send the packet to the control plane (or kernel) for logging. To emphasize this, a log action must also have a drop action.

Broadcom Range Checker Limitations

Broadcom platforms have only 24 range checkers. This is a separate resource from the total number of ACLs allowed. If you are creating a large ACL configuration, use port ranges for large ranges of more than 5 ports.

Inbound LOG Actions Only for Broadcom Switches

On Broadcom-based switches, LOG actions can only be done on inbound interfaces (the ingress direction), not on outbound interfaces (the egress direction).

SPAN Sessions that Reference an Outgoing Interface

SPAN sessions that reference an outgoing interface create mirrored packets based on the ingress interface before the routing/switching decision. For an example, see the SPAN Sessions that Reference an Outgoing Interface in the Network Troubleshooting chapter.

Tomahawk Hardware Limitations

Rate Limiting per Pipeline, Not Global

On Tomahawk switches, the field processor (FP) polices on a per-pipeline basis instead of globally, as with a Trident II switch. If packets come in to different switch ports that are on different pipelines on the ASIC, they might be rate limited differently.

For example, your switch is set so BFD is rate limited to 2000 packets per second. When the BFD packets are received on port1/pipe1 and port2/pipe2, they are each rate limited at 2000 pps; the switch is rate limiting at 4000 pps overall. Because there are four pipelines on a Tomahawk switch, you might see a fourfold increase of your configured rate limits.

Atomic Update Mode Enabled by Default

In Cumulus Linux, atomic update mode is enabled by default. If you have Tomahawk switches and plan to use SPAN and/or mangle rules, you must disable atomic update mode.

To do so, enable nonatomic update mode by setting the value for acl.non_atomic_update_mode to TRUE in /etc/cumulus/switchd.conf, then restart switchd.

acl.non_atomic_update_mode = TRUE

Packets Undercounted during ACL Updates

On Tomahawk switches, when updating egress FP rules, some packets do no get counted. This results in an underreporting of counts during ping-pong or incremental switchover.

Trident II+ Hardware Limitations

On a Trident II+ switch, the TCAM allocation for ACLs is limited to 2048 rules in atomic mode for a default setup instead of 4096, as advertised for ingress rules.

Trident3 Hardware Limitations

TCAM Allocation

On a Trident3 switch, the TCAM allocation for ACLs is limited to 2048 rules in atomic mode for a default setup instead of 4096, as advertised for ingress rules.

Enable Nonatomic Mode

On a Trident3 switch, you must enable nonatomic update mode before you can configure ERSPAN. To do so, set the value for acl.non_atomic_update_mode to TRUE in /etc/cumulus/switchd.conf, then restart switchd.

acl.non_atomic_update_mode = TRUE

Egress ACL Rules

On Trident3 switches, egress ACL rules matching on the output SVI interface match layer 3 routed packets only, not bridged packets. To match layer 2 traffic, use egress bridge member port-based rules.

iptables Interactions with cl-acltool

Because Cumulus Linux is a Linux operating system, the iptables commands can be used directly. However, consider using cl-acltool instead because:

For example, running the following command works:

cumulus@switch:~$ sudo iptables -A INPUT -p icmp --icmp-type echo-request -j DROP

And the rules appear when you run cl-acltool -L:

cumulus@switch:~$ sudo cl-acltool -L ip
-------------------------------
Listing rules of type iptables:
-------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 72 packets, 5236 bytes)
pkts bytes target  prot opt in   out   source    destination

0     0 DROP    icmp --  any  any   anywhere  anywhere      icmp echo-request

However, running cl-acltool -i or reboot removes them. To ensure all rules that can be in hardware are hardware accelerated, place them in the /etc/cumulus/acl/policy.conf file, then run cl-acltool -i.

Mellanox Spectrum Hardware Limitations

Due to hardware limitations in the Spectrum ASIC, BFD policers are shared between all BFD-related control plane rules. Specifically the following default rules share the same policer in the 00control_plan.rules file:

[iptables]
-A $INGRESS_CHAIN -p udp --dport $BFD_ECHO_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000
-A $INGRESS_CHAIN -p udp --dport $BFD_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000
-A $INGRESS_CHAIN -p udp --dport $BFD_MH_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000

[ip6tables]
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport $BFD_ECHO_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport $BFD_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7
-A $INGRESS_CHAIN --in-interface $INGRESS_INTF -p udp --dport $BFD_MH_PORT -j POLICE --set-mode pkt --set-rate 2000 --set-burst 2000 --set-class 7

To work around this limitation, set the rate and burst of all 6 of these rules to the same values, using the --set-rate and --set-burst options.

Where to Assign Rules

Generic Error Message Displayed after ACL Rule Installation Failure

After an ACL rule installation failure, a generic error message like the following is displayed:

cumulus@switch:$ sudo cl-acltool -i -p 00control_plane.rules
Using user provided rule file 00control_plane.rules
Reading rule file 00control_plane.rules ...
Processing rules in file 00control_plane.rules ...
error: hw sync failed (sync_acl hardware installation failed)
Installing acl policy... Rolling back ..
failed.

Dell S3048-ON Supports only 24K MAC Addresses

The Dell S3048-ON has a limit of 24576 MAC address entries instead of 32K for other 1G switches.

Mellanox Spectrum ASICs and INPUT Chain Rules

On switches with Mellanox Spectrum ASICs, INPUT chain rules are implemented using a trap mechanism. Packets headed to the CPU are assigned trap IDs. The default INPUT chain rules are mapped to these trap IDs. However, if a packet matches multiple traps, they are resolved by an internal priority mechanism that might be different from the rule priorities. Packets might not get policed by the default expected rule, but by another rule instead. For example, ICMP packets headed to the CPU are policed by the LOCAL rule instead of the ICMP rule. Also, multiple rules might share the same trap. In this case the policer that is applied is the largest of the policer values.

To work around this issue, create rules on the INPUT and FORWARD chains (INPUT,FORWARD).

Hardware Policing of Packets in the Input Chain

On certain platforms, there are limitations on hardware policing of packets in the INPUT chain. To work around these limitations, Cumulus Linux supports kernel based policing of these packets in software using limit/hashlimit matches. Rules with these matches are not hardware offloaded, but are ignored during hardware install.

ACLs Do not Match when the Output Port on the ACL is a Subinterface

Packets don’t get matched when a subinterface is configured as the output port. The ACL matches on packets only if the primary port is configured as an output port. If a subinterface is set as an output or egress port, the packets match correctly.

For example:

-A FORWARD --out-interface swp49s1.100 -j ACCEPT

Mellanox Switches and Egress ACL Matching on Bonds

On the Mellanox switch, ACL rules that match on an outbound bond interface are not supported. For example, the following rule is not supported:

[iptables]
-A FORWARD --out-interface <bond_intf> -j DROP

To work around this issue, duplicate the ACL rule on each physical port of the bond. For example:

[iptables]
-A FORWARD --out-interface <bond-member-port-1> -j DROP
-A FORWARD --out-interface <bond-member-port-2> -j DROP

Default Cumulus Linux ACL Configuration

The Cumulus Linux default ACL configuration is split into three parts: IP tables, IPv6 tables, and EB tables. The sections below describe the default configurations for each part. You can see the default file by clicking the Default ACL Configuration link:

Default ACL Configuration
cumulus@switch:~$ sudo cl-acltool -L all
-------------------------------
Listing rules of type iptables:
-------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 167 packets, 16481 bytes)
    pkts bytes target     prot opt in     out     source               destination
      0     0 DROP       all  --  swp+   any     240.0.0.0/5          anywhere
      0     0 DROP       all  --  swp+   any     loopback/8           anywhere
      0     0 DROP       all  --  swp+   any     base-address.mcast.net/8  anywhere
      0     0 DROP       all  --  swp+   any     255.255.255.255      anywhere
      0     0 SETCLASS   udp  --  swp+   any     anywhere             anywhere             udp dpt:3785 SETCLASS  class:7
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:3785 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   udp  --  swp+   any     anywhere             anywhere             udp dpt:3784 SETCLASS  class:7
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:3784 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   udp  --  swp+   any     anywhere             anywhere             udp dpt:4784 SETCLASS  class:7
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:4784 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   ospf --  swp+   any     anywhere             anywhere             SETCLASS  class:7
      0     0 POLICE     ospf --  any    any     anywhere             anywhere             POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   tcp  --  swp+   any     anywhere             anywhere             tcp dpt:bgp SETCLASS  class:7
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp dpt:bgp POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   tcp  --  swp+   any     anywhere             anywhere             tcp spt:bgp SETCLASS  class:7
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp spt:bgp POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   tcp  --  swp+   any     anywhere             anywhere             tcp dpt:5342 SETCLASS  class:7
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp dpt:5342 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   tcp  --  swp+   any     anywhere             anywhere             tcp spt:5342 SETCLASS  class:7
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp spt:5342 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   icmp --  swp+   any     anywhere             anywhere             SETCLASS  class:2
      1    84 POLICE     icmp --  any    any     anywhere             anywhere             POLICE  mode:pkt rate:100 burst:40
      0     0 SETCLASS   udp  --  swp+   any     anywhere             anywhere             udp dpts:bootps:bootpc SETCLASS  class:2
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:bootps POLICE  mode:pkt rate:100 burst:100
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:bootpc POLICE  mode:pkt rate:100 burst:100
      0     0 SETCLASS   tcp  --  swp+   any     anywhere             anywhere             tcp dpts:bootps:bootpc SETCLASS  class:2
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp dpt:bootps POLICE  mode:pkt rate:100 burst:100
      0     0 POLICE     tcp  --  any    any     anywhere             anywhere             tcp dpt:bootpc POLICE  mode:pkt rate:100 burst:100
      0     0 SETCLASS   udp  --  swp+   any     anywhere             anywhere             udp dpt:10001 SETCLASS  class:3
      0     0 POLICE     udp  --  any    any     anywhere             anywhere             udp dpt:10001 POLICE  mode:pkt rate:2000 burst:2000
      0     0 SETCLASS   igmp --  swp+   any     anywhere             anywhere             SETCLASS  class:6
      1    32 POLICE     igmp --  any    any     anywhere             anywhere             POLICE  mode:pkt rate:300 burst:100
      0     0 POLICE     all  --  swp+   any     anywhere             anywhere             ADDRTYPE match dst-type LOCAL POLICE  mode:pkt rate:1000 burst:1000 class:0
      0     0 POLICE     all  --  swp+   any     anywhere             anywhere             ADDRTYPE match dst-type IPROUTER POLICE  mode:pkt rate:400 burst:100 class:0
      0     0 SETCLASS   all  --  swp+   any     anywhere             anywhere             SETCLASS  class:0

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination
      0     0 DROP       all  --  swp+   any     240.0.0.0/5          anywhere
      0     0 DROP       all  --  swp+   any     loopback/8           anywhere
      0     0 DROP       all  --  swp+   any     base-address.mcast.net/8  anywhere
      0     0 DROP       all  --  swp+   any     255.255.255.255      anywhere

Chain OUTPUT (policy ACCEPT 107 packets, 12590 bytes)
    pkts bytes target     prot opt in     out     source               destination

TABLE mangle :
Chain PREROUTING (policy ACCEPT 172 packets, 17871 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain INPUT (policy ACCEPT 172 packets, 17871 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 111 packets, 18134 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain POSTROUTING (policy ACCEPT 111 packets, 18134 bytes)
    pkts bytes target     prot opt in     out     source               destination


TABLE raw :
Chain PREROUTING (policy ACCEPT 173 packets, 17923 bytes)
    pkts bytes target     prot opt in     out     source               destination

    Chain OUTPUT (policy ACCEPT 112 packets, 18978 bytes)
     pkts bytes target     prot opt in     out     source               destination


--------------------------------
Listing rules of type ip6tables:
--------------------------------
TABLE filter :
Chain INPUT (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target    prot opt in     out     source               destination
      0     0 DROP       all      swp+   any     ip6-mcastprefix/8    anywhere
      0     0 DROP       all      swp+   any     ::/128               anywhere
      0     0 DROP       all      swp+   any     ::ffff:0.0.0.0/96    anywhere
      0     0 DROP       all      swp+   any     localhost/128        anywhere
      0     0 POLICE     udp      swp+   any     anywhere             anywhere             udp dpt:3785 POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     udp      swp+   any     anywhere             anywhere             udp dpt:3784 POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     udp      swp+   any     anywhere             anywhere             udp dpt:4784 POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     ospf     swp+   any     anywhere             anywhere             POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     tcp      swp+   any     anywhere             anywhere             tcp dpt:bgp POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     tcp      swp+   any     anywhere             anywhere             tcp spt:bgp POLICE  mode:pkt rate:2000 burst:2000 class:7
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmp router-solicitation POLICE  mode:pkt rate:100 burst:100 class:2
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmp router-advertisement POLICE  mode:pkt rate:500 burst:500 class:2
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmp neighbour-solicitation POLICE  mode:pkt rate:400 burst:400 class:2
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmp neighbour-advertisement POLICE  mode:pkt rate:400 burst:400 class:2
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmptype 130 POLICE  mode:pkt rate:200 burst:100 class:6
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmptype 131 POLICE  mode:pkt rate:200 burst:100 class:6
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmptype 132 POLICE  mode:pkt rate:200 burst:100 class:6
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             ipv6-icmptype 143 POLICE  mode:pkt rate:200 burst:100 class:6
      0     0 POLICE     ipv6-icmp    swp+   any     anywhere             anywhere             POLICE  mode:pkt rate:64 burst:40 class:2
      0     0 POLICE     udp      swp+   any     anywhere             anywhere             udp dpts:dhcpv6-client:dhcpv6-server POLICE  mode:pkt rate:100 burst:100 class:2
      0     0 POLICE     tcp      swp+   any     anywhere             anywhere             tcp dpts:dhcpv6-client:dhcpv6-server POLICE  mode:pkt rate:100 burst:100 class:2
      0     0 POLICE     all      swp+   any     anywhere             anywhere             ADDRTYPE match dst-type LOCAL POLICE  mode:pkt rate:1000 burst:1000 class:0
      0     0 POLICE     all      swp+   any     anywhere             anywhere             ADDRTYPE match dst-type IPROUTER POLICE  mode:pkt rate:400 burst:100 class:0
      0     0 SETCLASS   all      swp+   any     anywhere             anywhere             SETCLASS  class:0

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target    prot opt in     out     source               destination
      0     0 DROP       all      swp+   any     ip6-mcastprefix/8    anywhere
      0     0 DROP       all      swp+   any     ::/128               anywhere
      0     0 DROP       all      swp+   any     ::ffff:0.0.0.0/96    anywhere
      0     0 DROP       all      swp+   any     localhost/128        anywhere

Chain OUTPUT (policy ACCEPT 5 packets, 408 bytes)
    pkts bytes target     prot opt in     out     source               destination


TABLE mangle :
Chain PREROUTING (policy ACCEPT 7 packets, 718 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain INPUT (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain POSTROUTING (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination


TABLE raw :
Chain PREROUTING (policy ACCEPT 7 packets, 718 bytes)
    pkts bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 0 packets, 0 bytes)
    pkts bytes target     prot opt in     out     source               destination

-------------------------------
Listing rules of type ebtables:
-------------------------------
TABLE filter :
Bridge table: filter

Bridge chain: INPUT, entries: 16, policy: ACCEPT
-d BGA -i swp+ -j setclass --class 7 , pcnt = 0 -- bcnt = 0
-d BGA -j police --set-mode pkt --set-rate 2000 --set-burst 2000 , pcnt = 0 -- bcnt = 0
-d 1:80:c2:0:0:2 -i swp+ -j setclass --class 7 , pcnt = 0 -- bcnt = 0
-d 1:80:c2:0:0:2 -j police --set-mode pkt --set-rate 2000 --set-burst 2000 , pcnt = 0 -- bcnt = 0
-d 1:80:c2:0:0:e -i swp+ -j setclass --class 6 , pcnt = 0 -- bcnt = 0
-d 1:80:c2:0:0:e -j police --set-mode pkt --set-rate 200 --set-burst 200 , pcnt = 0 -- bcnt = 0
-d 1:0:c:cc:cc:cc -i swp+ -j setclass --class 6 , pcnt = 0 -- bcnt = 0
-d 1:0:c:cc:cc:cc -j police --set-mode pkt --set-rate 200 --set-burst 200 , pcnt = 0 -- bcnt = 0
-p ARP -i swp+ -j setclass --class 2 , pcnt = 0 -- bcnt = 0
-p ARP -j police --set-mode pkt --set-rate 400 --set-burst 100 , pcnt = 0 -- bcnt = 0
-d 1:0:c:cc:cc:cd -i swp+ -j setclass --class 7 , pcnt = 0 -- bcnt = 0
-d 1:0:c:cc:cc:cd -j police --set-mode pkt --set-rate 2000 --set-burst 2000 , pcnt = 0 -- bcnt = 0
-p IPv4 -i swp+ -j ACCEPT , pcnt = 0 -- bcnt = 0
-p IPv6 -i swp+ -j ACCEPT , pcnt = 0 -- bcnt = 0
-i swp+ -j setclass --class 0 , pcnt = 0 -- bcnt = 0
-j police --set-mode pkt --set-rate 100 --set-burst 100 , pcnt = 0 -- bcnt = 0

Bridge chain: FORWARD, entries: 0, policy: ACCEPT

Bridge chain: OUTPUT, entries: 0, policy: ACCEPT

IP Tables

Action/Value Protocol/IP Address
Drop
Destination IP: Any
Source IPv4:
240.0.0.0/5
loopback/8
224.0.0.0/4
255.255.255.255
Set class: 7
Police: Packet rate 2000 burst 2000
Source IP: Any
Destination IP: Any
Protocol:
UDP/BFD Echo
UDP/BFD Control
UDP BFD Multihop Control
OSPF
TCP/BGP (spt dpt 179)
TCP/MLAG (spt dpt 5342)
Set Class: 6
Police: Rate 300 burst 100
Source IP: Any
Destination IP: Any
Protocol:
IGMP
Set class: 2
Police: Rate 100 burst 40
Source IP : Any
Destination IP: Any
Protocol:
ICMP
Set class: 2
Police: Rate 100 burst 100
Source IP: Any
Destination IP: Any
Protocol:
UDP/bootpc, bootps
Set class: 0
Police: Rate 1000 burst 1000
Source IP: Any
Destination IP: Any
ADDRTYPE match dst-type LOCAL
Note: LOCAL is any local address -> Receiving a packet with a destination matching a local IP address on the switch will go to the CPU.
Set class: 0
Police: Rate 400 burst 100
Source IP: Any
Destination IP: Any
ADDRTYPE match dst-type IPROUTER
Note: IPROUTER is any unresolved address -> On a l2/l3 boundary receiving a packet from L3 and needs to go to CPU in order to ARP for the destination.
Set class 0All

Set class is internal to the switch - it does not set any precedence bits.

IPv6 Tables

Action/Value Protocol/IP Address
DropSource IPv6:
ff00::/8
::
::ffff:0.0.0.0/96
localhost
Set class: 7
Police: Packet rate 2000 burst 2000
Source IPv6: Any
Destination IPv6: Any
Protocol:
UDP/BFD Echo
UDP/BFD Control
UDP BFD Multihop Control
OSPF
TCP/BGP (spt dpt 179)
Set class: 6
Police: Packet Rte: 200 burst 100
Source IPv6: Any
Destination IPv6: Any
Protocol:
Multicast Listener Query (MLD)
Multicast
Listener Report (MLD)
Multicast Listener Done (MLD
Multicast Listener Report V2
Set class: 2
Police: Packet rate: 100 burst 100
Source IPv6: Any
Destination IPv6: Any
Protocol:
ipv6-icmp router-solicitation
Set class: 2
Police: Packet rate: 500 burst 500
Source IPv6: Any
Destination IPv6: Any
Protocol:
ipv6-icmp router-advertisement POLICE
Set class: 2
Police: Packet rate: 400 burst 400
Source IPv6: Any
Destination IPv6: Any
Protocol:
ipv6-icmp neighbour-solicitation
ipv6-icmp neighbour-advertisement
Set class: 2
Police: Packet rate: 64 burst: 40
Source IPv6: Any
Destination IPv6: Any
Protocol:
Ipv6 icmp
Set class: 2
Police: Packet rate: 100 burst: 100
Source IPv6: Any
Destination IPv6: Any
Protocol:
UDP/dhcpv6-client:dhcpv6-server (Spts & dpts)
Police: Packet rate: 1000 burst 1000
Source IPv6: Any
Destination IPv6: Any
ADDRTYPE match dst-type LOCAL
Note: LOCAL is any local address -> Receiving a packet with a destination matching a local IPv6 address on the switch will go to the CPU.
Set class: 0
Police: Packet rate: 400 burst 100
ADDRTYPE match dst-type IPROUTER
Note: IPROUTER is an unresolved address -> On a l2/l3 boundary receiving a packet from L3 and needs to go to CPU in order to ARP for the destination.
Set class 0All

Set class is internal to the switch - it does not set any precedence bits.

EB Tables

Action/ValueProtocol/MAC Address
Set Class: 7
Police: packet rate: 2000 burst rate:2000
Any switchport input interface
BDPU
LACP=
Cisco PVST
Set Class: 6
Police: packet rate: 200 burst rate: 200
Any switchport input inteface
LLDP
CDP
Set Class: 2
Police: packet rate: 400 burst rate: 100
Any switchport input interface
ARP
Catch All:
Allow all traffic
Any switchport input interface
IPv4
IPv6
Catch All (applied at end):
Set class: 0
Police: packet rate 100 burst rate 100
Any switchport
ALL OTHER

Set class is internal to the switch. It does not set any precedence bits.

Filtering Learned MAC Addresses

On Broadcom switches, a MAC address is learned on a bridge regardless of whether or not a received packet is dropped by an ACL. This is due to how the hardware learns MAC addresses and occurs before the ACL lookup. This can be a security or resource problem as the MAC address table has the potential to get filled with bogus MAC addresses; a malfunctioning host, network error, loop, or malicious attack on a shared layer 2 platform can create an outage for other hosts if the same MAC address is learned on another port.

To prevent this from happening, Cumulus Linux filters frames before MAC learning occurs. Because MAC addresses and their port/VLAN associations are known at configuration time, you can create static MAC addresses, then create ingress ACLs to whitelist traffic from these MAC addresses and drop traffic otherwise.

This feature is specific to switches on the Broadcom platform only; on switches with Mellanox Spectrum ASICs, the input port ACL does not have these issues when learning MAC addresses.

Create a configuration similar to the following, where you associate a port and VLAN with a given MAC address, adding each one to the bridge:

cumulus@switch:~$ net add bridge bridge vids 100,200,300
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net add bridge bridge ports swp1-3
cumulus@switch:~$ net add bridge pre-up bridge fdb add 00:00:00:00:00:11 dev swp1 master static vlan 100
cumulus@switch:~$ net add bridge pre-up bridge fdb add 00:00:00:00:00:22 dev swp2 master static vlan 200
cumulus@switch:~$ net add bridge pre-up bridge fdb add 00:00:00:00:00:33 dev swp3 master static vlan 300
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3
    bridge-pvid 1
    bridge-vids 100 200 300
    bridge-vlan-aware yes
    pre-up bridge fdb add 00:00:00:00:00:11 dev swp1 master static vlan 100
    pre-up bridge fdb add 00:00:00:00:00:22 dev swp2 master static vlan 200
    pre-up bridge fdb add 00:00:00:00:00:33 dev swp3 master static vlan 300

If you need to list many MAC addresses, you can run a script to create the same configuration. For example, create a script called macs.txt and put in the bridge fdb add commands for each MAC address you need to configure:

cumulus@switch:~$ cat /etc/networks/macs.txt
#!/bin/bash
bridge fdb add 00:00:00:00:00:11 dev swp1 master static vlan 100
bridge fdb add 00:00:00:00:00:22 dev swp2 master static vlan 200
bridge fdb add 00:00:00:00:00:33 dev swp3 master static vlan 300
bridge fdb add 00:00:00:00:00:44 dev swp4 master static vlan 400
bridge fdb add 00:00:00:00:00:55 dev swp5 master static vlan 500
bridge fdb add 00:00:00:00:00:66 dev swp6 master static vlan 600

Then create the configuration using NCLU:

cumulus@switch:~$ net add bridge bridge vids 100,200,300
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net add bridge bridge ports swp1-3
cumulus@switch:~$ net add bridge pre-up /etc/networks/macs.txt
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4

auto swp5
iface swp5

auto swp6
iface swp6

auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3 swp4 swp5 swp6
    bridge-pvid 1
    bridge-vids 100 200 300
    bridge-vlan-aware yes
    pre-up bridge fdb add 00:00:00:00:00:11 dev swp1 master static vlan 100
    pre-up bridge fdb add 00:00:00:00:00:22 dev swp2 master static vlan 200
    pre-up bridge fdb add 00:00:00:00:00:33 dev swp3 master static vlan 300
    pre-up bridge fdb add 00:00:00:00:00:44 dev swp4 master static vlan 400
    pre-up bridge fdb add 00:00:00:00:00:55 dev swp5 master static vlan 500
    pre-up bridge fdb add 00:00:00:00:00:66 dev swp6 master static vlan 600

Interactions with EVPN

If you are using EVPN, local static MAC addresses added to the local FDB are exported as static MAC addresses to remote switches. Remote MAC addresses are added as MAC addresses to the remote FDB.

Services and Daemons in Cumulus Linux

Services (also known as daemons) and processes are at the heart of how a Linux system functions. Most of the time, a service takes care of itself; you just enable and start it, then let it run. However, because a Cumulus Linux switch is a Linux system, you can dig deeper if you like. Services can start multiple processes as they run. Services are important to monitor on a Cumulus Linux switch.

You manage services in Cumulus Linux in the following ways:

systemd and the systemctl Command

In general, you manage services using systemd via the systemctl command. You use it with any service on the switch to start, stop, restart, reload, enable, disable, reenable, or get the status of the service.

cumulus@switch:~$ sudo systemctl start | stop | restart | status | reload | enable | disable | reenable SERVICENAME.service

For example to restart networking, run the command:

cumulus@switch:~$ sudo systemctl restart networking.service

The service name is written after the systemctl subcommand, not before it.

To show all the services currently running, run the systemctl status command. For example:

cumulus@switch:~$ sudo systemctl status
● switch
    State: running
      Jobs: 0 queued
    Failed: 0 units
    Since: Thu 2019-01-10 00:19:34 UTC; 23h ago
    CGroup: /
            ├─init.scope
            │ └─1 /sbin/init
            └─system.slice
              ├─haveged.service
              │ └─234 /usr/sbin/haveged --Foreground --verbose=1 -w 1024
              ├─sysmonitor.service
              │ ├─  658 /bin/bash /usr/lib/cumulus/sysmonitor
              │ └─26543 sleep 60
              ├─systemd-udevd.service
              │ └─218 /lib/systemd/systemd-udevd
              ├─system-ntp.slice
              │ └─ntp@mgmt.service
              │   └─vrf
              │     └─mgmt
              │       └─12108 /usr/sbin/ntpd -n -u ntp:ntp -g
              ├─cron.service
              │ └─274 /usr/sbin/cron -f -L 38
              ├─system-serial\x2dgetty.slice
              │ └─serial-getty@ttyS0.service
              │   └─745 /sbin/agetty -o -p -- \u --keep-baud 115200,38400,9600 ttyS0 vt220
              ├─nginx.service
              │ ├─332 nginx: master process /usr/sbin/nginx -g daemon on; master_process on;
              │ └─333 nginx: worker process
              ├─auditd.service
              │ └─235 /sbin/auditd
              ├─rasdaemon.service
              │ └─275 /usr/sbin/rasdaemon -f -r
              ├─clagd.service
              │ └─11443 /usr/bin/python /usr/sbin/clagd --daemon 169.254.1.2 peerlink.4094 44:39:39:ff:40:9
              --priority 100 --vxlanAnycas
              ├─switchd.service
              │ └─430 /usr/sbin/switchd -vx
              ...

systemctl Subcommands

systemctl has a number of subcommands that perform a specific operation on a given service.

There is often little reason to interact with the services directly using these commands. If a critical service crashes or encounters an error, it is automatically respawned by systemd. systemd is effectively the caretaker of services in modern Linux systems and is responsible for starting all the necessary services at boot time.

Ensure a Service Starts after Multiple Restarts

By default, systemd is configured to try to restart a particular service only a certain number of times within a given interval before the service fails to start at all. The settings, StartLimitInterval (which defaults to 10 seconds) and StartBurstLimit (which defaults to 5 attempts) are stored in the service script; however, many services override these defaults, sometimes with much longer times. For example, switchd.service sets StartLimitInterval=10m and StartBurstLimit=3; therefore, if you restart switchd more than 3 times in 10 minutes, it does not start.

When the restart fails for this reason, you see a message similar to the following:

Job for switchd.service failed. See 'systemctl status switchd.service' and 'journalctl -xn' for details.

systemctl status switchd.service shows output similar to:

Active: failed (Result: start-limit) since Thu 2016-04-07 21:55:14 UTC; 15s ago

To clear this error, run systemctl reset-failed switchd.service. If you know you are going to restart frequently (multiple times within the StartLimitInterval), you can run the same command before you issue the restart request. This also applies to stop followed by start.

Keep systemd Services from Hanging after Starting

If you start, restart, or reload any systemd service that can be started from another systemd service, you must use the --no-block option with systemctl. Otherwise, that service or even the switch itself might hang after starting or restarting.

Identify Active Listener Ports for IPv4 and IPv6

You can identify the active listener ports under both IPv4 and IPv6 using the netstat command:

cumulus@switch:~$ netstat -nlp --inet --inet6
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name
tcp        0      0 0.0.0.0:53              0.0.0.0:*               LISTEN      444/dnsmasq
tcp        0      0 0.0.0.0:22              0.0.0.0:*               LISTEN      874/sshd
tcp6       0      0 :::53                   :::*                    LISTEN      444/dnsmasq
tcp6       0      0 :::22                   :::*                    LISTEN      874/sshd
udp        0      0 0.0.0.0:28450           0.0.0.0:*                           839/dhclient
udp        0      0 0.0.0.0:53              0.0.0.0:*                           444/dnsmasq
udp        0      0 0.0.0.0:68              0.0.0.0:*                           839/dhclient
udp        0      0 192.168.0.42:123        0.0.0.0:*                           907/ntpd
udp        0      0 127.0.0.1:123           0.0.0.0:*                           907/ntpd
udp        0      0 0.0.0.0:123             0.0.0.0:*                           907/ntpd
udp        0      0 0.0.0.0:4784            0.0.0.0:*                           909/ptmd
udp        0      0 0.0.0.0:3784            0.0.0.0:*                           909/ptmd
udp        0      0 0.0.0.0:3785            0.0.0.0:*                           909/ptmd
udp6       0      0 :::58352                :::*                                839/dhclient
udp6       0      0 :::53                   :::*                                444/dnsmasq
udp6       0      0 fe80::a200:ff:fe00::123 :::*                                907/ntpd
udp6       0      0 ::1:123                 :::*                                907/ntpd
udp6       0      0 :::123                  :::*                                907/ntpd
udp6       0      0 :::4784                 :::*                                909/ptmd
udp6       0      0 :::3784                 :::*                                909/ptmd

Identify Services Currently Active or Stopped

To determine which services are currently active or stopped, run the cl-service-summary command:

cumulus@switch:~$ cl-service-summary
Service cron               enabled    active
Service ssh                enabled    active
Service syslog             enabled    active
Service asic-monitor       enabled    inactive
Service clagd              enabled    inactive
Service cumulus-poe                   inactive
Service lldpd              enabled    active
Service mstpd              enabled    active
Service neighmgrd          enabled    active
Service netd               enabled    active
Service netq-agent         enabled    active
Service ntp                enabled    active
Service portwd             enabled    active
Service ptmd               enabled    active
Service pwmd               enabled    active
Service smond              enabled    active
Service switchd            enabled    active
Service sysmonitor         enabled    active
Service rdnbrd             disabled   inactive
Service frr                enabled    inactive
...

You can also run the systemctl list-unit-files --type service command to list all services on the switch and see which ones are enabled:

cumulus@switch:~$ systemctl list-unit-files --type service
UNIT FILE                              STATE
aclinit.service                        enabled
acltool.service                        enabled
acpid.service                          disabled
asic-monitor.service                   enabled
auditd.service                         enabled
autovt@.service                        disabled
bmcd.service                           disabled
bootlog.service                        enabled
bootlogd.service                       masked  
bootlogs.service                       masked  
bootmisc.service                       masked  
checkfs.service                        masked  
checkroot-bootclean.service            masked  
checkroot.service                      masked
clagd.service                          enabled
console-getty.service                  disabled
console-shell.service                  disabled
container-getty@.service               static  
cron.service                           enabled
cryptdisks-early.service               masked  
cryptdisks.service                     masked  
cumulus-aclcheck.service               static  
cumulus-core.service                   static  
cumulus-fastfailover.service           enabled
cumulus-firstboot.service              disabled
cumulus-hyperconverged.service         disabled
cumulus-platform.service               enabled  
...

Identify Essential Services

If you need to know which services are required to run when the switch boots, run:

cumulus@switch:~$ systemctl list-dependencies --before basic.target

To see which services are needed for networking, run:

cumulus@switch:~$ systemctl list-dependencies --after network.target
   ├─switchd.service
   ├─wd_keepalive.service
   └─network-pre.target

To identify the services needed for a multi-user environment, run:

cumulus@switch:~$ systemctl list-dependencies --before multi-user.target

 ●  ├─bootlog.service
   ├─systemd-readahead-done.service
   ├─systemd-readahead-done.timer
   ├─systemd-update-utmp-runlevel.service
   └─graphical.target
   └─systemd-update-utmp-runlevel.service

Important Services

The following table lists the most important services in Cumulus Linux.

Service NameDescriptionAffects Forwarding?
switchdHardware abstraction daemon. Synchronizes the kernel with the ASIC.YES
sx_sdkInterfaces with the Spectrum ASIC. Only on Spectrum switches.YES
portwdReads pluggable information over the I2C bus. Identifies and classifies the optics that are inserted into the system. Sets interface speeds and capabilities to match the optics.YES, eventually, if optics are added or removed
frrFRRouting. Handles routing protocols. There are separate processes for each routing protocol, such as bgpd and ospfd.YES if routing
clagCumulus link aggregation daemon. Handles MLAG.YES if using MLAG
neighmgrdSynchronizes MAC address information if using MLAG.YES if using MLAG
mstpdSpanning tree protocol daemon.YES if using layer 2
ptmdPrescriptive Topology Manager. Verifies cabling based on LLDP output. Also sets up BFD sessions.YES if using BFD
netdNCLU back end.NO
rsyslogHandles logging of syslog messages.NO
ntpNetwork time protocol.NO
ledmgrdLED manager. Reads the state of system LEDs.NO
sysmonitorWatches and logs critical system load (free memory, disk, CPU).NO
lldpdHandles Tx/Rx of LLDP information.NO
smondReads platform sensors and fan information from pwmd.NO
pwmdReads and sets fan speeds.NO

Configuring switchd

switchd is the daemon at the heart of Cumulus Linux. It communicates between the switch and Cumulus Linux, and all the applications running on Cumulus Linux.

The switchd configuration is stored in /etc/cumulus/switchd.conf.

The switchd File System

switchd also exports a file system, mounted on /cumulus/switchd, that presents all the switchd configuration options as a series of files arranged in a tree structure. To show the contents, run the tree /cumulus/switchd command. The following example shows output for a switch with one switch port configured:

cumulus@switch:~$ sudo tree /cumulus/switchd/
/cumulus/switchd/
├── clear
│   └── stats
│       ├── vlan
│       └── vxlan
├── config
│   ├── acl
│   │   ├── flow_based_mirroring
│   │   ├── non_atomic_update_mode
│   │   ├── optimize_hw
│   │   └── vxlan_tnl_arp_punt_disable
│   ├── arp
│   │   ├── drop_during_failed_state
│   │   └── next_hops
│   ├── bridge
│   │   ├── broadcast_frame_to_cpu
│   │   └── optimized_mcast_flood
│   ├── buf_util
│   │   ├── measure_interval
│   │   └── poll_interval
│   ├── coalesce
│   │   ├── offset
│   │   ├── reducer
│   │   └── timeout
│   ├── disable_internal_hw_err_restart
│   ├── disable_internal_parity_restart
│   ├── hal
│   │   └── bcm
│   │       ├── l3
│   │       │   └── per_vlan_router_mac_lookup_for_vrrp
│   │       ├── linkscan_interval
│   │       ├── logging
│   │       │   └── l3mc
│   │       ├── per_vlan_router_mac_lookup
│   │       └── vxlan_support
│   ├── ignore_non_swps
│   ├── interface
│   │   ├── swp1
│   │   │   ├── ethtool_mode
│   │   │   ├── interface_mode
│   │   │   ├── port_security
│   │   │   │   ├── enable
│   │   │   │   ├── mac_limit
│   │   │   │   ├── static_mac
│   │   │   │   ├── sticky_aging
│   │   │   │   ├── sticky_mac
│   │   │   │   ├── sticky_timeout
│   │   │   │   ├── violation_mode
│   │   │   │   └── violation_timeout
│   │   │   └── storm_control
│   │   │       ├── broadcast
│   │   │       ├── multicast
│   │   │       └── unknown_unicast
...

Configure switchd Parameters

To configure the switchd parameters, edit the /etc/cumulus/switchd.conf file. An example is provided below.

cumulus@switch:~$ sudo nano /etc/cumulus/switchd.conf
#
# /etc/cumulus/switchd.conf - switchd configuration file
#

# Statistic poll interval (in msec)
#stats.poll_interval = 2000

# Buffer utilization poll interval (in msec), 0 means disable
#buf_util.poll_interval = 0

# Buffer utilization measurement interval (in mins)
#buf_util.measure_interval = 0

# Optimize ACL HW resources for better utilization
#acl.optimize_hw = FALSE

# Enable Flow based mirroring.
#acl.flow_based_mirroring = TRUE

# Enable non atomic acl update
acl.non_atomic_update_mode = FALSE

# Send ARPs for next hops
#arp.next_hops = TRUE

# Kernel routing table ID, range 1 - 2^31, default 254
#route.table = 254
...

When you update the /etc/cumulus/switchd.conf file, you must restart switchd for the changes to take effect. See Restart switchd, below.

Restart switchd

Whenever you modify a switchd hardware configuration file (for example, you update any *.conf file that requires making a change to the switching hardware, like /etc/cumulus/datapath/traffic.conf), you must restart the switchd service for the change to take effect:

cumulus@switch:~$ sudo systemctl restart switchd.service

You do not have to restart the switchd service when you update a network interface configuration (for example, when you edit the /etc/network/interfaces file).

Restarting the switchd service causes all network ports to reset in addition to resetting the switch hardware configuration.

Power over Ethernet - PoE

Cumulus Linux supports Power over Ethernet (PoE) and PoE+, so certain Cumulus Linux switches can supply power from Ethernet switch ports to enabled devices over the Ethernet cables that connect them. PoE is capable of powering devices up to 15W, while PoE+ can power devices up to 30W. Configuration for power negotiation is done over LLDP.

The currently supported platforms include:

PoE Basics

PoE functionality is provided by the cumulus-poe package. When a powered device is connected to the switch via an Ethernet cable:

Power is available as follows:

PSU 1PSU 2PoE Power Budget
920Wx750W
x920W750W
920W920W1650W

The AS4610-54P has an LED on the front panel to indicate PoE status:

Link state and PoE state are completely independent of each other. When a link is brought down on a particular port using ip link <port> down, power on that port is not turned off; however, LLDP negotiation is not possible.

Configure PoE

You use the poectl command utility to configure PoE on a switch that supports the feature. You can:

The PoE configuration resides in /etc/cumulus/poe.conf. The file lists all the switch ports, whether PoE is enabled for those ports and the priority for each port.

Sample poe.conf file ...
[enable]
swp1 = enable
swp2 = enable
swp3 = enable
swp4 = enable
swp5 = enable
swp6 = enable
swp7 = enable
swp8 = enable
swp9 = enable
swp10 = enable
swp11 = enable
swp12 = enable
swp13 = enable
swp14 = enable
swp15 = enable
swp16 = enable
swp17 = enable
swp18 = enable
swp19 = enable
swp20 = enable
swp21 = enable
swp22 = enable
swp23 = enable
swp24 = enable
swp25 = enable
swp26 = enable
swp27 = enable
swp28 = enable
swp29 = enable
swp30 = enable
swp31 = enable
swp32 = enable
swp33 = enable
swp34 = enable
swp35 = enable
swp36 = enable
swp37 = enable
swp38 = enable
swp39 = enable
swp40 = enable
swp41 = enable
swp42 = enable
swp43 = enable
swp44 = enable
swp45 = enable
swp46 = enable
swp47 = enable
swp48 = enable

[priority]
swp1 = low
swp2 = low
swp3 = low
swp4 = low
swp5 = low
swp6 = low
swp7 = low
swp8 = low
swp9 = low
swp10 = low
swp11 = low
swp12 = low
swp13 = low
swp14 = low
swp15 = low
swp16 = low
swp17 = low
swp18 = low
swp19 = low
swp20 = low
swp21 = low
swp22 = low
swp23 = low
swp24 = low
swp25 = low
swp26 = low
swp27 = low
swp28 = low
swp29 = low
swp30 = low
swp31 = low
swp32 = low
swp33 = low
swp34 = low
swp35 = low
swp36 = low
swp37 = low
swp38 = low
swp39 = low
swp40 = low
swp41 = low
swp42 = low
swp43 = low
swp44 = low
swp45 = low
swp46 = low
swp47 = low
swp48 = low

By default, PoE and PoE+ are enabled on all Ethernet/1G switch ports, and these ports are set with a low priority. Switch ports can have low, high or critical priority.

There is no additional configuration for PoE+.

To change the priority for one or more switch ports, run poectl -p swp# [low|high|critical]. For example:

cumulus@switch:~$ sudo poectl -p swp1-swp5,swp7 high

To disable PoE for one or more ports, run poectl -d [port_numbers]:

cumulus@switch:~$ sudo poectl -d swp1-swp5,swp7

To display PoE information for a set of switch ports, run poectl -i [port_numbers]:

cumulus@switch:~$ sudo poectl -i swp10-swp13
Port          Status            Allocated    Priority  PD type      PD class   Voltage   Current    Power
-----   --------------------   -----------   -------- -----------   --------   -------   -------   ---------
swp10   connected              negotiating   low      IEEE802.3at   4          53.5 V     25 mA    3.9 W
swp11   searching              n/a           low      IEEE802.3at   none        0.0 V      0 mA    0.0 W
swp12   connected              n/a           low      IEEE802.3at   2          53.5 V     25 mA    1.4 W
swp13   connected              51.0 W        low      IEEE802.3at   4          53.6 V     72 mA    3.8 W

The Status can be one of the following:

The Allocated column displays how much PoE power has been allocated to the port, which can be one of the following:

To see all the PoE information for a switch, run poectl -s:

cumulus@switch:~$ poectl -s
System power:
  Total:      730.0 W
  Used:        11.0 W
  Available:  719.0 W
Connected ports:
  swp11, swp24, swp27, swp48

The set commands (priority, enable, disable) either succeed silently or display an error message if the command fails.

The poectl command takes the following arguments:

Argument Description
-h, --helpShow this help message and exit
-i, --port-info
<port-list>
Returns detailed information for the specified ports. You can specify a range of ports. For example: -i swp1-swp5,swp10.
Note: On an Edge-Core AS4610-54P switch, the voltage reported by the poectl -i command and measured through a power meter connected to the device varies by 5V. The current and power readings are correct and no difference is seen for them.
-a, --allReturns PoE status and detailed information for all ports.
-p, --priority
<port-list> <priority>
Sets priority for the specified ports: low, high, critical.
-d, --disable-ports <port-list>Disables PoE operation on the specified ports.
-e, --enable-ports <port-list>Enables PoE operation on the specified ports.
-s, --systemReturns PoE status for the entire switch.
-r, --reset <port-list>Performs a hardware reset on the specified ports. Use this if one or more ports are stuck in an error state. This does not reset any configuration settings for the specified ports.
-v, --versionDisplays version information.
-j, --jsonDisplays output in JSON format.
--saveSaves the current configuration. The saved configuration is automatically loaded on system boot.
--loadLoads and applies the saved configuration.

Troubleshooting

You can troubleshoot PoE and PoE+ using the following utilities and files:

LLDP requires network connectivity, so verify that the link is up.

cumulus@switch:~$ net show interface swp20
    Name    MAC                Speed      MTU  Mode
--  ------  -----------------  -------  -----  ---------
UP  swp20   44:38:39:00:00:04  1G        1500  Access/L2

View LLDP Information Using lldpcli

You can run lldpcli to view the LLDP information that has been received on a switch port. For example:

cumulus@switch:~$ sudo lldpcli show neighbors ports swp20 protocol lldp hidden details
-------------------------------------------------------------------------------
LLDP neighbors:
-------------------------------------------------------------------------------
Interface:    swp20, via: LLDP, RID: 2, Time: 0 day, 00:03:34
  Chassis:
    ChassisID:    mac 68:c9:0b:25:54:7c
    SysName:      ihm-ubuntu
    SysDescr:     Ubuntu 14.04.2 LTS Linux 3.14.4+ #1 SMP Thu Jun 26 00:54:44 UTC 2014 armv7l
    MgmtIP:       fe80::6ac9:bff:fe25:547c
    Capability:   Bridge, off
    Capability:   Router, off
    Capability:   Wlan, off
    Capability:   Station, on
  Port:
    PortID:       mac 68:c9:0b:25:54:7c
    PortDescr:    eth0
    PMD autoneg:  supported: yes, enabled: yes
      Adv:          10Base-T, HD: yes, FD: yes
      Adv:          100Base-TX, HD: yes, FD: yes
      MAU oper type: 100BaseTXFD - 2 pair category 5 UTP, full duplex mode
    MDI Power:    supported: yes, enabled: yes, pair control: no
      Device type:  PD
      Power pairs:  spare
      Class:        class 4
      Power type:   2
      Power Source: Primary power source
      Power Priority: low
      PD requested power Value: 51000
      PSE allocated power Value: 51000
  UnknownTLVs: 
    TLV:          OUI: 00,01,42, SubType: 1, Len: 1 05
    TLV:          OUI: 00,01,42, SubType: 1, Len: 1 0D
-------------------------------------------------------------------------------

View LLDP Information Using tcpdump

You can use tcpdump to view the LLDP frames being transmitted and received. For example:

cumulus@switch:~$ sudo tcpdump -v -v -i swp20 ether proto 0x88cc
tcpdump: listening on swp20, link-type EN10MB (Ethernet), capture size 262144 bytes
18:41:47.559022 LLDP, length 211
    Chassis ID TLV (1), length 7
      Subtype MAC address (4): 00:30:ab:f2:d7:a5 (oui Unknown)
      0x0000:  0400 30ab f2d7 a5
    Port ID TLV (2), length 6
      Subtype Interface Name (5): swp20
      0x0000:  0573 7770 3230
    Time to Live TLV (3), length 2: TTL 120s
      0x0000:  0078
    System Name TLV (5), length 13: dni-3048up-09
      0x0000:  646e 692d 3330 3438 7570 2d30 39
    System Description TLV (6), length 68
      Cumulus Linux version 3.0.1~1466303042.2265c10 running on dni 3048up
      0x0000:  4375 6d75 6c75 7320 4c69 6e75 7820 7665
      0x0010:  7273 696f 6e20 332e 302e 317e 3134 3636
      0x0020:  3330 3330 3432 2e32 3236 3563 3130 2072
      0x0030:  756e 6e69 6e67 206f 6e20 646e 6920 3330
      0x0040:  3438 7570
    System Capabilities TLV (7), length 4
      System  Capabilities [Bridge, Router] (0x0014)
      Enabled Capabilities [Router] (0x0010)
      0x0000:  0014 0010
    Management Address TLV (8), length 12
      Management Address length 5, AFI IPv4 (1): 10.0.3.190
      Interface Index Interface Numbering (2): 2
      0x0000:  0501 0a00 03be 0200 0000 0200
    Management Address TLV (8), length 24
      Management Address length 17, AFI IPv6 (2): fe80::230:abff:fef2:d7a5
      Interface Index Interface Numbering (2): 2
      0x0000:  1102 fe80 0000 0000 0000 0230 abff fef2
      0x0010:  d7a5 0200 0000 0200
    Port Description TLV (4), length 5: swp20
      0x0000:  7377 7032 30
    Organization specific TLV (127), length 9: OUI IEEE 802.3 Private (0x00120f)
      Link aggregation Subtype (3)
        aggregation status [supported], aggregation port ID 0
      0x0000:  0012 0f03 0100 0000 00
    Organization specific TLV (127), length 9: OUI IEEE 802.3 Private (0x00120f)
      MAC/PHY configuration/status Subtype (1)
        autonegotiation [supported, enabled] (0x03)
        PMD autoneg capability [10BASE-T fdx, 100BASE-TX fdx, 1000BASE-T fdx] (0x2401)
        MAU type 100BASEFX fdx (0x0012)
      0x0000:  0012 0f01 0324 0100 12
    Organization specific TLV (127), length 12: OUI IEEE 802.3 Private (0x00120f)
      Power via MDI Subtype (2)
        MDI power support [PSE, supported, enabled], power pair spare, power class class4
      0x0000:  0012 0f02 0702 0513 01fe 01fe
    Organization specific TLV (127), length 5: OUI Unknown (0x000142)
      0x0000:  0001 4201 0d
    Organization specific TLV (127), length 5: OUI Unknown (0x000142)
      0x0000:  0001 4201 01
    End TLV (0), length 0

Log poed Events in syslog

The poed service logs the following events to syslog when:

Configuring a Global Proxy

You configure global HTTP and HTTPS proxies in the /etc/profile.d/ directory of Cumulus Linux. To do so, set the http_proxy and https_proxy variables, which tells the switch the address of the proxy server to use to fetch URLs on the command line. This is useful for programs such as apt/apt-get, curl and wget, which can all use this proxy.

  1. In a terminal, create a new file in the /etc/profile.d/ directory. In the code example below, the file is called proxy.sh, and is created using the text editor nano.
cumulus@switch:~$ sudo nano /etc/profile.d/proxy.sh
  1. Add a line to the file to configure either an HTTP or an HTTPS proxy, or both.
http_proxy=http://myproxy.domain.com:8080
export http_proxy

https_proxy=https://myproxy.domain.com:8080
export https_proxy
  1. Create a file in the /etc/apt/apt.conf.d directory and add the following lines to the file for acquiring the HTTP and HTTPS proxies; the example below uses http_proxy as the file name:
cumulus@switch:~$ sudo nano /etc/apt/apt.conf.d/http_proxy
Acquire::http::Proxy "http://myproxy.domain.com:8080";
Acquire::https::Proxy "https://myproxy.domain.com:8080";
  1. Add the proxy addresses to /etc/wgetrc; you may have to uncomment the http_proxy and https_proxy lines:
cumulus@switch:~$ sudo nano /etc/wgetrc
...
https_proxy = https://myproxy.domain.com:8080
http_proxy = http://myproxy.domain.com:8080
...
  1. Run the source command, to execute the file in the current environment:
cumulus@switch:~$ source /etc/profile.d/proxy.sh

The proxy is now configured. The echo command can be used to confirm aproxy is set up correctly:

cumulus@switch:~$ echo $http_proxy
http://myproxy.domain.com:8080
cumulus@switch:~$ echo $https_proxy
https://myproxy.domain.com:8080
Setting up an apt package cache

HTTP API

Cumulus Linux implements an HTTP application programing interface to OpenStack ML2 driver and NCLU. Instead of accessing Cumulus Linux using SSH, you can interact with the switch using an HTTP client, such as cURL, HTTPie or a web browser.

HTTP API Basics

The supporting software for the API is installed with Cumulus Linux.

To enable the HTTP API service, run the following systemd command:

cumulus@switch:~$ sudo systemctl enable restserver

Use the systemctl start and systemctl stop commands to start or stop the HTTP API service:

cumulus@switch:~$ sudo systemctl start restserver
cumulus@switch:~$ sudo systemctl stop restserver

Each service runs as a background daemon.

Configuration

You configure the HTTP API services in the /etc/nginx/sites-available/nginx-restapi.conf configuration file.

IP and Port Settings

You can modify the IP:port combinations to which services listen by changing the parameters of the listen directive(s). By default, nginx-restapi.conf has only one listen parameter.

All URLs must use HTTPS instead of HTTP.

For more information on the listen directive, refer to the NGINX documentation.

Security

Authentication

The default configuration requires all HTTP requests from external sources (not internal switch traffic) to set the HTTP Basic Authentication header.

The user and password must correspond to a user on the host switch.

Transport Layer Security

All traffic must be secured in transport using TLSv1.2 by default. Cumulus Linux contains a self-signed certificate and private key used server-side in this application so that it works out of the box, but Cumulus Networks recommends you use your own certificates and keys. Certificates must be in the PEM format.

For step by step documentation for generating self-signed certificates and keys, and installing them to the switch, refer to the Ubuntu Certificates and Security documentation.

Do not copy the cumulus.pem or cumulus.key files. After installation, edit the ssl_certificate and ssl_certificate_key values in the configuration file for your hardware.

cURL Examples

This section includes several example cURL commands you can use to send HTTP requests to a host. The following settings are used for these examples:

Requests for NCLU require setting the Content-Type request header to be set to application/json.

The cURL -k flag is necessary when the server uses a self-signed certificate. This is the default configuration (see the Security section). To display the response headers, include the -D flag in the command.

To retrieve a list of all available HTTP endpoints:

cumulus@switch:~$ curl -X GET -k -u user:pw https://192.168.0.32:8080

To run net show counters on the host as a remote procedure call:

cumulus@switch:~$ curl -X POST -k -u user:pw -H "Content-Type: application/json" -d '{"cmd": "show counters"}' https://192.168.0.32:8080/nclu/v1/rpc

To add a bridge using ML2:

cumulus@switch:~$ curl -X PUT -k -u user:pw https://192.168.0.32:8080/ml2/v1/bridge/"br1"/200

Caveats

The /etc/restapi.conf file is not listed in the net show configuration files command output.

Layer 1 and Switch Ports

This section discusses how to configure network interfaces and DHCP delays and servers. The Prescriptive Topology Manager (PTM) cabling verification tool is also discussed.

Interface Configuration and Management

ifupdown is the network interface manager for Cumulus Linux. Cumulus Linux uses an updated version of this tool, ifupdown2.

For more information on network interfaces, see Switch Port Attributes.

By default, ifupdown is quiet. Use the verbose option (-v) to show commands as they are executed when bringing an interface down or up.

Basic Commands

To bring up the physical connection to an interface or apply changes to an existing interface, run the sudo ifup <interface> command. The following example command brings up the physical connection to swp1:

cumulus@switch:~$ sudo ifup swp1

To bring down the physical connection to a single interface, run the sudo ifdown <interface> command. The following example command brings down the physical connection to swp1:

cumulus@switch:~$ sudo ifdown swp1

The ifdown command always deletes logical interfaces after bringing them down. When you bring down the physical connection to an interface, it is brought back up automatically after any future reboots or configuration changes with ifreload -a.

To administratively bring the interface up or down; for example, to bring down a port, bridge, or bond but not the physical connection for a port, bridge, or bond, you can use the the --admin-state option. Alternatively, you can use NCLU commands.

When you put an interface into an admin down state, the interface remains down after any future reboots or configuration changes with ifreload -a.

NCLU Commands

To put an interface into an admin down state, run the net add interface <interface> link down command.

cumulus@switch:~$ net add interface swp1 link down
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto swp1
iface swp1
    link-down yes

To bring the interface back up, run the net del interface <interface> link down command.

cumulus@switch:~$ net del interface swp1 link down
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

To put an interface into an admin down state, run the sudo ifdown <interface> --admin-state command:

cumulus@switch:~$ sudo ifdown swp1 --admin-state

These commands create the following configuration in the /etc/network/interfaces file:

auto swp1
iface swp1
    link-down yes

To bring the interface back up, run the sudo ifup <interface> --admin-state command:

cumulus@switch:~$ sudo ifup swp1 --admin-state

To see the link and administrative state, use the ip link show command. In the following example, swp1 is administratively UP and the physical link is UP (LOWER_UP flag).

cumulus@switch:~$ ip link show dev swp1
3: swp1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff

For additional information on interface administrative state and physical state, refer to this knowledge base article.

ifupdown2 Interface Classes

ifupdown2 enables you to group interfaces into separate classes, where a class is a user-defined label that groups interfaces that share a common function (such as uplink, downlink or compute). You specify classes in the /etc/network/interfaces file.

The most common class is auto, which you configure like this:

auto swp1
iface swp1

You can add other classes using the allow prefix. For example, if you have multiple interfaces used for uplinks, you can define a class called uplinks:

auto swp1
allow-uplink swp1
iface swp1 inet static
    address 10.1.1.1/31

auto swp2
allow-uplink swp2
iface swp2 inet static
    address 10.1.1.3/31

This allows you to perform operations on only these interfaces using the --allow=uplinks option. You can still use the -a options because these interfaces are also in the auto class:

cumulus@switch:~$ sudo ifup --allow=uplinks
cumulus@switch:~$ sudo ifreload -a

If you are using Management VRF, you can use the special interface class called mgmt and put the management interface into that class. The management VRF must have an IPv6 address in addition to an IPv4 address to work correctly.

The mgmt interface class is not supported with NCLU commands.

allow-mgmt eth0
iface eth0 inet dhcp
    vrf mgmt

allow-mgmt mgmt
iface mgmt
    address 127.0.0.1/8
    address ::1/128
    vrf-table auto

All ifupdown2 commands (ifup, ifdown, ifquery, ifreload) can take a class. Include the --allow=<class> option when you run the command. For example, to reload the configuration for the management interface described above, run:

cumulus@switch:~$ sudo ifreload --allow=mgmt

Use the -a option to bring up or down all interfaces that are marked with the common auto class in the /etc/network/interfaces file.

To administratively bring up all interfaces marked auto, run:

cumulus@switch:~$ sudo ifup -a

To administratively bring down all interfaces marked auto, run:

cumulus@switch:~$ sudo ifdown -a

To reload all network interfaces marked auto, use the ifreload command. This command is equivalent to running ifdown then ifup; however, ifreload skips unchanged configurations:

cumulus@switch:~$ sudo ifreload -a

Certain syntax checks are done by default. As a precaution, apply configurations only if the syntax check passes. Use the following compound command:

cumulus@switch:~$ sudo bash -c "ifreload -s -a && ifreload -a"

For more information, see the individual man pages for ifup(8), ifdown(8), ifreload(8).

Configure a Loopback Interface

Cumulus Linux has a loopback interface preconfigured in the /etc/network/interfaces file. When the switch boots up, it has a loopback interface called lo, which is up and assigned an IP address of 127.0.0.1.

The loopback interface lo must always be specified in the /etc/network/interfaces file and must always be up.

ifupdown Behavior with Child Interfaces

By default, ifupdown recognizes and uses any interface present on the system that is listed as a dependent of an interface (for example, a VLAN, bond, or physical interface). You are not required to list interfaces in the interfaces file unless they need a specific configuration for MTU, link speed, and so on. If you need to delete a child interface, delete all references to that interface from the interfaces file.

In the following example, swp1 and swp2 do not need an entry in the interfaces file. The following stanzas defined in /etc/network/interfaces provide the exact same configuration:

With Child Interfaces Defined:

auto swp1
iface swp1

auto swp2
iface swp2

auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge-ports swp1 swp2
    bridge-vids 1-100
    bridge-pvid 1
    bridge-stp on

Without Child Interfaces Defined

auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge-ports swp1 swp2
    bridge-vids 1-100
    bridge-pvid 1
    bridge-stp on</code></pre></td>

In the following example, swp1.100 and swp2.100 do not need an entry in the interfaces file. The following stanzas defined in /etc/network/interfaces provide the exact same configuration:

With Child Interfaces Defined

auto swp1.100
iface swp1.100

auto swp2.100
iface swp2.100

auto br-100
iface br-100
    address 10.0.12.2/24
    address 2001:dad:beef::3/64
    bridge-ports swp1.100 swp2.100
    bridge-stp on

Without Child Interfaces Defined

auto br-100
iface br-100
    address 10.0.12.2/24
    address 2001:dad:beef::3/64
    bridge-ports swp1.100 swp2.100
    bridge-stp on

For more information about bridges in traditional mode and bridges in VLAN-aware mode, read this knowledge base article.

ifupdown2 Interface Dependencies

ifupdown2 understands interface dependency relationships. When you run ifup and ifdown with all interfaces, the commands always run with all interfaces in dependency order. When you run ifup and ifdown with the interface list on the command line, the default behavior is to not run with dependents; however, if there are any built-in dependents, they will be brought up or down.

To run with dependents when you specify the interface list, use the --with-depends option. The --with-depends option walks through all dependents in the dependency tree rooted at the interface you specify. Consider the following example configuration:

auto bond1
iface bond1
    address 100.0.0.2/16
    bond-slaves swp29 swp30

auto bond2
iface bond2
    address 100.0.0.5/16
    bond-slaves swp31 swp32

auto br2001
iface br2001
    address 12.0.1.3/24
    bridge-ports bond1.2001 bond2.2001
    bridge-stp on

The ifup --with-depends br2001 command brings up all dependents of br2001: bond1.2001, bond2.2001, bond1, bond2, bond1.2001, bond2.2001, swp29, swp30, swp31, swp32.

cumulus@switch:~$ sudo ifup --with-depends br2001

The ifdown --with-depends br2001 command brings down all dependents of br2001: bond1.2001, bond2.2001, bond1, bond2, bond1.2001, bond2.2001, swp29, swp30, swp31, swp32.

cumulus@switch:~$ sudo ifdown --with-depends br2001

ifdown2 always deletes logical interfaces after bringing them down. Use the --admin-state option if you only want to administratively bring the interface up or down. In the above example, ifdown br2001 deletes br2001.

To guide you through which interfaces will be brought down and up, use the --print-dependency option.

For example, run ifquery --print-dependency=list -a to show the dependency list for all interfaces:

cumulus@switch:~$ sudo ifquery --print-dependency=list -a
lo : None
eth0 : None
bond0 : ['swp25', 'swp26']
bond1 : ['swp29', 'swp30']
bond2 : ['swp31', 'swp32']
br0 : ['bond1', 'bond2']
bond1.2000 : ['bond1']
bond2.2000 : ['bond2']
br2000 : ['bond1.2000', 'bond2.2000']
bond1.2001 : ['bond1']
bond2.2001 : ['bond2']
br2001 : ['bond1.2001', 'bond2.2001']
swp40 : None
swp25 : None
swp26 : None
swp29 : None
swp30 : None
swp31 : None
swp32 : None

To print the dependency list of a single interface, run the ifquery --print-dependency=list <interface> command. The following example command shows the dependency list for br2001:

cumulus@switch:~$ sudo ifquery --print-dependency=list br2001
br2001 : ['bond1.2001', 'bond2.2001']
bond1.2001 : ['bond1']
bond2.2001 : ['bond2']
bond1 : ['swp29', 'swp30']
bond2 : ['swp31', 'swp32']
swp29 : None
swp30 : None
swp31 : None
swp32 : None

To show the dependency information for an interface in dot format, run the ifquery --print-dependency=dot <interface> command. The following example command shows the dependency information for interface br2001 in dot format:

cumulus@switch:~$ sudo ifquery --print-dependency=dot br2001
/* Generated by GvGen v.0.9 (http://software.inl.fr/trac/wiki/GvGen) */
digraph G {
    compound=true;
    node1 [label="br2001"];
    node2 [label="bond1.2001"];
    node3 [label="bond2.2001"];
    node4 [label="bond1"];
    node5 [label="bond2"];
    node6 [label="swp29"];
    node7 [label="swp30"];
    node8 [label="swp31"];
    node9 [label="swp32"];
    node1->node2;
    node1->node3;
    node2->node4;
    node3->node5;
    node4->node6;
    node4->node7;
    node5->node8;
    node5->node9;
}

You can use dot to render the graph on an external system where dot is installed.

To print the dependency information of the entire interfaces file, run the following command:

cumulus@switch:~$ sudo ifquery --print-dependency=dot -a >interfaces_all.dot

Subinterfaces

On Linux, an interface is a network device that can be either physical, like a switch port (for example, swp1) or virtual, like a VLAN (for example, vlan100). A VLAN subinterface is a VLAN device on an interface, and the VLAN ID is appended to the parent interface using dot (.) VLAN notation. For example, a VLAN with ID 100 that is a subinterface of swp1 is named swp1.100. The dot VLAN notation for a VLAN device name is a standard way to specify a VLAN device on Linux. Many Linux configuration tools, such as ifupdown2 and its predecessor ifupdown, recognize such a name as a VLAN interface name.

A VLAN subinterface only receives traffic tagged for that VLAN; therefore, swp1.100 only receives packets tagged with VLAN 100 on switch port swp1. Similarly, any packets transmitted from swp1.100 are tagged with VLAN 100.

In an MLAG configuration, the peer link interface that connects the two switches in the MLAG pair has a VLAN subinterface named 4094 by default if you configured the subinterface with NCLU. The peerlink.4094 subinterface only receives traffic tagged for VLAN 4094.

ifup and Upper (Parent) Interfaces

When you run ifup on a logical interface (like a bridge, bond or VLAN interface), if the ifup results in the creation of the logical interface, it implicitly tries to execute on the interface’s upper (or parent) interfaces as well.

Consider this example configuration:

auto br100
iface br100
    bridge-ports bond1.100 bond2.100

auto bond1
iface bond1
    bond-slaves swp1 swp2

If you run ifdown bond1, ifdown deletes bond1 and the VLAN interface on bond1 (bond1.100); it also removes bond1 from the bridge br100. Next, when you run ifup bond1, it creates bond1 and the VLAN interface on bond1 (bond1.100); it also executes ifup br100 to add the bond VLAN interface (bond1.100) to the bridge br100.

There can be cases where an upper interface (like br100) is not in the right state, which can result in warnings. The warnings are mostly harmless.

If you want to disable these warnings, you can disable the implicit upper interface handling by setting skip_upperifaces=1 in the /etc/network/ifupdown2/ifupdown2.conf file.

With skip_upperifaces=1, you have to explicitly execute ifup on the upper interfaces. In this case, you will have to run ifup br100 after an ifup bond1 to add bond1 back to bridge br100.

Although specifying a subinterface like swp1.100 and then running ifup swp1.100 results in the automatic creation of the swp1 interface in the kernel, Cumulus Networks recommends you specify the parent interface swp1 as well. A parent interface is one where any physical layer configuration can reside, such as link-speed 1000 or link-duplex full. If you only create swp1.100 and not swp1, then you cannot run ifup swp1 because you did not specify it.

Configure IP Addresses

To configure IP addresses, run the following commands.

NCLU Commands

The following commands configure three IP addresses for swp1: two IPv4 addresses, and one IPv6 address.

cumulus@switch:~$ net add interface swp1 ip address 12.0.0.1/30
cumulus@switch:~$ net add interface swp1 ip address 12.0.0.2/30
cumulus@switch:~$ net add interface swp1 ipv6 address 2001:DB8::1/126
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following code snippet in the /etc/network/interfaces file:

auto swp1
iface swp1
    address 12.0.0.1/30
    address 12.0.0.2/30
    address 2001:DB8::1/126

You can specify both IPv4 and IPv6 addresses for the same interface.

For IPv6 addresses, you can create or modify the IP address for an interface using either :: or 0:0:0 notation. Both of the following examples are valid:

cumulus@switch:~$ net add bgp neighbor 2620:149:43:c109:0:0:0:5 remote-as internal
cumulus@switch:~$ net add interface swp1 ipv6 address 2001:DB8::1/126

NCLU adds the address method and address family when needed, specifically when you are creating DHCP or loopback interfaces.

auto lo
iface lo inet loopback

Linux Commands

In the /etc/network/interfaces file, list all IP addresses under the iface section. The following command example adds IP address 10.0.0.1/30 and 10.0.0.2/30 to swp1.

auto swp1
iface swp1
    address 10.0.0.1/30
    address 10.0.0.2/30

The address method and address family are not mandatory; they default to inet/inet6 and static. However, you must specify inet/inet6 when you are creating DHCP or loopback interfaces.

auto lo
iface lo inet loopback

You can specify both IPv4 and IPv6 addresses in the same iface stanza:

auto swp1
iface swp1
    address 192.0.2.1/30
    address 192.0.2.2/30
    address 2001:DB8::1/126

A runtime configuration is non-persistent, which means the configurationyou create here does not persist after you reboot the switch.

To make non-persistent changes to interfaces at runtime, use ip addr add:

cumulus@switch:~$ sudo ip addr add 192.0.2.1/30 dev swp1
cumulus@switch:~$ sudo ip addr add 2001:DB8::1/126 dev swp1

To remove an addresses from an interface, use ip addr del:

cumulus@switch:~$ sudo ip addr del 192.0.2.1/30 dev swp1
cumulus@switch:~$ sudo ip addr del 2001:DB8::1/126 dev swp1

For more details on the options available to manage and query interfaces, see man ip.

To show the assigned IP address on an interface, run the ip addr show command. The following example command shows the assigned IP address on swp1.

cumulus@switch:~$ ip addr show dev swp1
3: swp1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
    inet 192.0.2.1/30 scope global swp1
    inet 192.0.2.2/30 scope global swp1
    inet6 2001:DB8::1/126 scope global tentative
        valid_lft forever preferred_lft forever

Specify IP Address Scope

ifupdown2 does not honor the configured IP address scope setting in the /etc/network/interfaces file, treating all addresses as global. It does not report an error. Consider this example configuration:

auto swp2
iface swp2
    address 35.21.30.5/30
    address 3101:21:20::31/80
    scope link

When you run ifreload -a on this configuration, ifupdown2 considers all IP addresses as global.

cumulus@switch:~$ ip addr show swp2
5: swp2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 74:e6:e2:f5:62:82 brd ff:ff:ff:ff:ff:ff
inet 35.21.30.5/30 scope global swp2
valid_lft forever preferred_lft forever
inet6 3101:21:20::31/80 scope global 
valid_lft forever preferred_lft forever
inet6 fe80::76e6:e2ff:fef5:6282/64 scope link 
valid_lft forever preferred_lft forever

To work around this issue, configure the IP address scope:

NCLU Commands

Run the following commands:

cumulus@switch:~$ net add interface swp6 post-up ip address add 71.21.21.20/32 dev swp6 scope site
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following code snippet in the /etc/network/interfaces file:

auto swp6
iface swp6
    post-up ip address add 71.21.21.20/32 dev swp6 scope site
Linux Commands

In the /etc/network/interfaces file, configure the IP address scope using post-up ip address add <address> dev <interface> scope <scope>. For example:

auto swp6
iface swp6
    post-up ip address add 71.21.21.20/32 dev swp6 scope site

Then run the ifreload -a command on this configuration.

The following configuration shows the correct scope:

cumulus@switch:~$ ip addr show swp6
9: swp6: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 74:e6:e2:f5:62:86 brd ff:ff:ff:ff:ff:ff
inet 71.21.21.20/32 scope site swp6
valid_lft forever preferred_lft forever
inet6 fe80::76e6:e2ff:fef5:6286/64 scope link
valid_lft forever preferred_lft forever

Purge Existing IP Addresses on an Interface

By default, ifupdown2 purges existing IP addresses on an interface. If you have other processes that manage IP addresses for an interface, you can disable this feature.

NCLU Commands

To disable IP address purge on an interface, run the following commands:

cumulus@switch:~$ net add interface swp1 address-purge no
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration snippet in the /etc/network/interfaces file:

auto swp1
iface swp1
    address-purge no
Linux Commands

In the /etc/network/interfaces file, add address-purge no to the interface configuration. The following example command disables IP address purge on swp1.

cumulus@switch:~# sudo nano /etc/network/interfaces

auto swp1
iface swp1
    address-purge no

Purging existing addresses on interfaces with multiple iface stanzas is not supported. Doing so can result in the configuration of multiple addresses for an interface after you change an interface address and reload the configuration with ifreload -a. If this happens, you must shut down and restart the interface with ifup and ifdown, or manually delete superfluous addresses with ip address delete specify.ip.address.here/mask dev DEVICE. See also the Caveats and Errata section below for cautions about using multiple iface stanzas for the same interface.

Specify User Commands

You can specify additional user commands in the /etc/network/interfaces file. The interface stanzas in /etc/network/interfaces can have a command that runs at pre-up, up, post-up, pre-down, down, and post-down:

NCLU commands

To add a command to an interface stanza, run the following commands:

cumulus@switch:~$ net add interface swp1 post-up /sbin/foo bar
cumulus@switch:~$ net add interface ip address 12.0.0.1/30
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto swp1
iface swp1
    address 12.0.0.1/30
    post-up /sbin/foo bar

If your post-up command also starts, restarts, or reloads any systemd service, you must use the --no-block option with systemctl. Otherwise, that service or even the switch itself might hang after starting or restarting. For example, to restart the dhcrelay service after bringing up VLAN 100, first run:

cumulus@switch:~$ net add vlan 100 post-up systemctl --no-block restart dhcrelay.service

This command creates the following configuration in the /etc/network/interfaces file:

auto bridge
iface bridge
    bridge-vids 100
    bridge-vlan-aware yes

auto vlan100
iface vlan100
    post-up systemctl --no-block restart dhcrelay.service
    vlan-id 100
    vlan-raw-device bridge

Linux Commands

To add a command to an interface stanza, add the command in the /etc/network/interfaces file. For example:

cumulus@switch:~# sudo nano /etc/network/interfaces

auto swp1
iface swp1
    address 12.0.0.1/30
    up /sbin/foo bar

If your post-up command also starts, restarts, or reloads any systemd service, you must use the --no-block option with systemctl. Otherwise, that service or even the switch itself might hang after starting or restarting. For example, to restart the dhcrelay service after bringing up a VLAN, the /etc network/interfaces configuration looks like this:

auto bridge.100
iface bridge.100 
    post-up systemctl --no-block restart dhcrelay.service

You can add any valid command in the sequence to bring an interface up or down; however, limit the scope to network-related commands associated with the particular interface. For example, it does not make sense to install a Debian package on ifup of swp1, even though it is technically possible. See man interfaces for more details.

Source Interface File Snippets

Sourcing interface files helps organize and manage the interfaces file. For example:

cumulus@switch:~$ sudo cat /etc/network/interfaces
# The loopback network interface
auto lo
iface lo inet loopback

# The primary network interface
auto eth0
iface eth0 inet dhcp

source /etc/network/interfaces.d/bond0

The contents of the sourced file used above are:

cumulus@switch:~$ sudo cat /etc/network/interfaces.d/bond0
auto bond0
iface bond0
    address 14.0.0.9/30
    address 2001:ded:beef:2::1/64
    bond-slaves swp25 swp26

Use Globs for Port Lists

Globs define a range of ports.

NCLU Commands

NCLU supports globs to define port lists (a range of ports). You must use commas to separate different ranges of ports in the NCLU command; for example:

cumulus@switch:~$ net add bridge bridge ports swp1-4,6,10-12
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands produce the following snippet in the /etc/network/interfaces file. The file renders the list of ports individually.

...

auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3 swp4 swp6 swp10 swp11 swp12
    bridge-vlan-aware yes
auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4

auto swp6
iface swp6

auto swp10
iface swp10

auto swp11
iface swp11

auto swp12
iface swp12
Linux Commands

Use the glob keyword to specify bridge ports and bond slaves:

auto br0
iface br0
    bridge-ports glob swp1-6.100

auto br1
iface br1
    bridge-ports glob swp7-9.100  swp11.100 glob swp15-18.100

Mako Templates

ifupdown2 supports Mako-style templates. The Mako template engine is run over the interfaces file before parsing.

While ifupdown2 supports Mako templates, NCLU does not understand them. As a result, NCLU cannot read or write to the /etc/network/interfaces file.

Use the template to declare cookie-cutter bridges in the interfaces file:

And use it to declare addresses in the interfaces file:

%for i in [1,12]:
auto swp${i}
iface swp${i}
    address 10.20.${i}.3/24

In Mako syntax, use square brackets ([1,12]) to specify a list of individual numbers (in this case, 1 and 12). Use range(1,12) to specify a range of interfaces.

You can test your template and confirm it evaluates correctly by running mako-render /etc/network/interfaces.

To comment out content in Mako templates, use double hash marks (##). For example:

## % for i in range(1, 4):
## auto swp${i}
## iface swp${i}
## % endfor
##

For more examples of configuring Mako templates, read this knowledge base article.

Run ifupdown Scripts under /etc/network/ with ifupdown2

Unlike the traditional ifupdown system, ifupdown2 does not run scripts installed in /etc/network/*/ automatically to configure network interfaces.

To enable or disable ifupdown2 scripting, edit the addon_scripts_support line in the /etc/network/ifupdown2/ifupdown2.conf file. 1 enables scripting and 2 disables scripting. The following example enables scripting.

cumulus@switch:~$ sudo nano /etc/network/ifupdown2/ifupdown2.conf
# Support executing of ifupdown style scripts.
# Note that by default python addon modules override scripts with the same name
addon_scripts_support=1

ifupdown2 sets the following environment variables when executing commands:

Add Descriptions to Interfaces

You can add descriptions to interfaces configured in the /etc/network/interfaces file by using the alias keyword.

NCLU Commands

The following commands create an alias for swp1:

cumulus@switch:~$ net add interface swp1 alias hypervisor_port_1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following code snippet:

auto swp1
iface swp1
    alias hypervisor_port_1
Linux Commands

In the /etc/network/interfaces file, add a description using the alias keyword:

cumulus@switch:~# sudo nano /etc/network/interfaces

auto swp1
iface swp1
    alias swp1 hypervisor_port_1

You can query the interface description.

NCLU Commands

To show the description (alias) for an interface, run the net show interface <interface> command. The following example command shows the description for swp1:

cumulus@switch$ net show interface swp1
    Name   MAC                Speed     MTU   Mode
--  ----   -----------------  -------   -----  ---------
UP  swp1   44:38:39:00:00:04  1G        1500   Access/L2
Alias
-----
hypervisor_port_1

To show the interface description (alias) for all interfaces on the switch, run the net show interface alias command. For example:

cumulus@switch:~$ net show interface alias
State    Name            Mode              Alias
-----    -------------   -------------     ------------------
UP       bond01          LACP
UP       bond02          LACP
UP       bridge          Bridge/L2
UP       eth0            Mgmt
UP       lo              Loopback          loopback interface
UP       mgmt            Interface/L3
UP       peerlink        LACP
UP       peerlink.4094   SubInt/L3
UP       swp1            BondMember        hypervisor_port_1
UP       swp2            BondMember        to Server02
...

To show the interface description for all interfaces on the switch in JSON format, run the net show interface alias json command.

Linux Commands

To show the description (alias) for an interface, run the ip link show command. The alias appears on the alias line:

cumulus@switch$ ip link show swp1
3: swp1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT qlen 500
    link/ether aa:aa:aa:aa:aa:bc brd ff:ff:ff:ff:ff:ff
    alias hypervisor_port_1

Interface descriptions also appear in the SNMP OID IF-MIB::ifAlias.

  • Aliases are limited to 256 characters.
  • Avoid using apostrophes or non-ASCII characters in the alias string. Cumulus Linux does not parse these characters.

Caveats and Errata

Even though ifupdown2 supports the inclusion of multiple iface stanzas for the same interface, Cumulus Networks recommends that you use a single iface stanza for each interface. If you must specify more than one iface stanza; for example, if the configuration for a single interface comes from many places, like a template or a sourced file, make sure the stanzas do not specify the same interface attributes. Otherwise, unexpected behavior can result.

In the following example, swp1 is configured in two places: the /etc/network/interfaces file and the /etc/network/interfaces.d/speed_settings file. ifupdown2 correctly parses this configuration because the same attributes are not specified in multiple iface stanzas.

cumulus@switch:~$ sudo cat /etc/network/interfaces

source /etc/network/interfaces.d/speed_settings

auto swp1
iface swp1
  address 10.0.14.2/24

cumulus@switch:~$ cat /etc/network/interfaces.d/speed_settings

auto swp1
iface swp1
  link-speed 1000
  link-duplex full

You cannot purge existing addresses on interfaces with multiple iface stanzas.

ifupdown2 and sysctl

For sysctl commands in the pre-up, up, post-up, pre-down, down, and post-down lines that use the $IFACE variable, if the interface name contains a dot (.), ifupdown2 does not change the name to work with sysctl. For example, the interface name bridge.1 is not converted to bridge/1.

Long Interface Names

Interface names can be a maximum of 15 characters in length and you cannot use a number as the first character. Longer interface names might result in errors. To resolve long interface name issues, remove the interface from the /etc/network/interfaces file, then restart networking.service.

cumulus@switch:~$ sudo nano /etc/network/interfaces
cumulus@switch:~$ sudo systemctl restart networking.service

Switch Port Attributes

Cumulus Linux exposes network interfaces for several types of physical and logical devices:

Each physical network interface (port) has a number of configurable settings:

Most of these settings are configured automatically for you, depending upon your switch ASIC; however, you must always set MTU manually.

For Spectrum ASICs, MTU is the only port attribute you can directly configure. The Spectrum firmware configures FEC, link speed, duplex mode and auto-negotiation automatically, following a predefined list of parameter settings until the link comes up. However, you can disable FEC if necessary, which forces the firmware to not try any FEC options.

For Broadcom-based switches, Cumulus Networks recommends that you enable auto-negotiation on each port. When enabled, Cumulus Linux automatically configures the best link parameter settings based on the module type (speed, duplex, auto-negotiation, and FEC, where supported).

This topic describes the auto-negotiation, link speed, duplex mode, MTU, and FEC settings and provides a table showing the default configuration for various port and cable types. Breakout port configuration, logical switch port limitations, and troubleshooting is also provided.

Auto-negotiation

By default on a Broadcom-based switch, auto-negotiation is disabled - except on 10G and 1000BASE-T fixed copper switch ports, where it is required for links to work. For RJ-45 SFP adapters, you need to manually configure the desired link speed and auto-negotiation as described in the default settings table below.

If you disable auto-negotiation later or never enable it, then you have to configure any settings that deviate from the port default - such as duplex mode, FEC, and link speed settings.

Some module types support auto-negotiation while others do not. To enable a simpler configuration, Cumulus Linux allows you to configure auto-negotiation on all port types on Broadcom switches; the port configuration software then configures the underlying hardware according to its capabilities.

If you do decide to disable auto-negotiation, be aware of the following:

  • You must manually set any non-default link speed, duplex, pause, and FEC.
  • Disabling auto-negotiation on a 1G optical cable prevents detection of single fiber breaks.
  • You cannot disable auto-negotiation on 1GT or 10GT fixed copper switch ports.

For 1000BASE-T RJ-45 SFP adapters, auto-negotiation is automatically done on the SFP PHY, so enabling auto-negotiation on the port settings is not required. You must manually configure these ports using the settings below.

Depending upon the connector used for a port, enabling auto-negotiation also enables forward error correction (FEC), if the cable requires it (see the table below). The correct FEC mode is set based on the speed of the cable when auto-negotiation is enabled.

To configure auto-negotiation for a switch:

NCLU Commands

Run the net add interface <interface> link autoneg command. The following example commands enable auto-negotiation for the swp1 interface:

cumulus@switch:~$ net add interface swp1 link autoneg on
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands
  1. Edit the /etc/network/interfaces file. The following example disables auto-negotiation for the swp1 interface.
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
    link-autoneg off
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

You can use ethtool to configure auto-negotiation. The following example command enables auto-negotiation for the swp1 interface:

ethtool -s swp1 speed 10000 duplex full autoneg on|off

A runtime configuration is non-persistent; the configuration you create here does not persist after you reboot the switch.

Any time you enable auto-negotiation, Cumulus Linux restores the default configuration settings specified in the table below.

Port Speed and Duplex Mode

Cumulus Linux supports both half- and full-duplex configurations. Half-duplex is supported only with speeds of less than 1G.

Supported port speeds include 100M, 1G, 10G, 25G, 40G, 50G and 100G. In Cumulus Linux, you set the speed on Broadcom-based switch in Mbps, where the setting for 1G is 1000, 40G is 40000, and 100G is 100000.

You can configure ports to one speed less than their maximum speed.

Switch Port TypeLowest Configurable Speed
1G100 Mb
10G1 Gigabit (1000 Mb)
40G10G*
100G50G* & 40G (with or without breakout port), 25G*, 10G*

*Requires the port to be converted into a breakout port. See Configure Breakout Ports below.

Platform Limitations

  • On Lenovo NE2572O switches, swp1 through swp8 only support 25G speed.
  • For 10G and 1G SFPs inserted in a 25G port on a Broadcom platform, you must edit the /etc/cumulus/ports.conf file and configure the four ports in the same core to be 10G. See Caveats and Errata.

To configure the port speed and duplex mode:

NCLU Commands

Run the net add interface <interface> link speed command. The following commands configure the port speed for the swp1 interface. The duplex mode setting defaults to full. You only need to specify link duplex if you want to set half-duplex mode.

cumulus@switch:~$ net add interface swp1 link speed 10000
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The above commands create the following /etc/network/interfaces file code snippet:

auto swp1
iface swp1
    link-speed 10000

The following commands configure the port speed and set half-duplex mode for the swp31 interface.

cumulus@switch:~$ net add interface swp31 link speed 100 
cumulus@switch:~$ net add interface swp31 link duplex half
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The above commands create the following /etc/network/interfaces file code snippet:

auto swp31
iface swp31
    link-speed 100
    link-duplex half
Linux Commands

Edit the /etc/network/interfaces file to create a persistent configuration for the port speeds:

  1. Add the appropriate lines for each switch port stanza. The following example shows that the port speed for the swp1 interface is set to 10G and the duplex mode is set to full.

    If you specify the port speed in the /etc/network/interfaces file, you must also specify the duplex mode setting; otherwise, the interface defaults to half duplex.

cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
    address 10.1.1.1/24
    link-speed 10000
    link-duplex full
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

You can use ethtool to configure the port speed and duplex mode for your switch ports. You must specify both the port speed and the duplex mode in the ethtool command; auto-negotiation is optional.

The following example command sets the port speed to 10G and duplex mode to full on the swp1 interface:

cumulus@switch:~$  ethtool -s swp1 speed 10000 duplex full

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

MTU

Interface MTU applies to traffic traversing the management port, front panel/switch ports, bridge, VLAN subinterfaces, and bonds (both physical and logical interfaces). MTU is the only interface setting that you must set manually.

On Mellanox switches, the default MTU setting is 9238 in Cumulus Linux. On Broadcom switches, the default MTU setting is 1500. To change the setting, run the following commands:

NCLU Commands

Run the net add interface <interface> mtu command. The following example command sets MTU to 9000 for the swp1 interface.

cumulus@switch:~$ net add interface swp1 mtu 9000
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following code snippet:

auto swp1
iface swp1
    mtu 9000
Linux Commands
  1. Edit the /etc/network/interfaces file. The following example sets MTU to 9000 for the swp1 interface.
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
    mtu 9000
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

Run the ip link set command. The following example command sets the swp1 interface to Jumbo Frame MTU=9000.

cumulus@switch:~$ sudo ip link set dev swp1 mtu 9000

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

Some switches might not support the same maximum MTU setting in hardware for both the management interface (eth0) and the data plane ports.

Set a Policy for Global System MTU

For a global policy to set MTU, create a policy document (called mtu.json). For example:

cumulus@switch:~$ sudo cat /etc/network/ifupdown2/policy.d/mtu.json
{
  "address": {"defaults": { "mtu": "9216" }
            }
}

If your platform does not support a high MTU on eth0, you can set a lower MTU with the following command:

cumulus@switch:~$ net add interface eth0 mtu 1500
cumulus@switch:~$ net commit

The policies and attributes in any file in /etc/network/ifupdown2/policy.d/ override the default policies and attributes in /var/lib/ifupdown2/policy.d/.

MTU for a Bridge

The MTU setting is the lowest MTU of any interface that is a member of the bridge (every interface specified in bridge-ports in the bridge configuration of the /etc/network/interfaces file). There is no need to specify an MTU on the bridge. Consider this bridge configuration:

auto bridge
iface bridge
    bridge-ports bond1 bond2 bond3 bond4 peer5
    bridge-vids 100-110
    bridge-vlan-aware yes

For bridge to have an MTU of 9000, set the MTU for each of the member interfaces (bond1 to bond 4, and peer5), to 9000 at minimum.

Use MTU 9216 for a bridge

Two common MTUs for jumbo frames are 9216 and 9000 bytes. The corresponding MTUs for the VNIs would be 9166 and 8950.

When configuring MTU for a bond, configure the MTU value directly under the bond interface; the configured value is inherited by member links/slave interfaces. If you need a different MTU on the bond, set it on the bond interface, as this ensures the slave interfaces pick it up. There is no need to specify MTU on the slave interfaces.

VLAN interfaces inherit their MTU settings from their physical devices or their lower interface; for example, swp1.100 inherits its MTU setting from swp1. Therefore, specifying an MTU on swp1 ensures that swp1.100 inherits the MTU setting for swp1.

If you are working with VXLANs, the MTU for a virtual network interface (VNI must be 50 bytes smaller than the MTU of the physical interfaces on the switch, as those 50 bytes are required for various headers and other data. Also, consider setting the MTU much higher than 1500.

The MTU for an SVI interface, such as vlan100, is derived from the bridge. When you use NCLU to change the MTU for an SVI and the MTU setting is higher than it is for the other bridge member interfaces, the MTU for all bridge member interfaces changes to the new setting. If you need to use a mixed MTU configuration for SVIs, (if some SVIs have a higher MTU and some lower), set the MTU for all member interfaces to the maximum value, then set the MTU on the specific SVIs that need to run at a lower MTU.

To show the MTU setting for an interface:

NCLU Commands

Run the net show interface <interface> command:

cumulus@switch:~$ net show interface swp1
    Name    MAC                Speed      MTU  Mode
--  ------  -----------------  -------  -----  ---------
UP  swp1    44:38:39:00:00:04  1G        1500  Access/L2
Linux Commands

Run the ip link show <interface> command:

cumulus@switch:~$ ip link show dev swp1
3: swp1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT qlen 500
   link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff

Bring Down an Interface for a Bridge Member

When you bring down an interface for a bridge member, the MTU for the interface and the MTU for the bridge are both set to the default value of 1500 for Broadcom switches and 9238 for Mellanox switches. To work around this, run ifdown on the interface, then run the sudo ip link set dev <interface> mtu command.

For example:

sudo ifdown swp3
sudo ip link set dev swp3 mtu 9192

As an alternative, add a post-down command in the /etc/network/interfaces file to reset the MTU of the interface. For example:

auto swp3
iface swp3
    bridge-vids 106 109 119 141 150-151
    mtu 9192
    post-down /sbin/ip link set dev swp3 mtu 9192

FEC

Forward Error Correction (FEC) is an encoding and decoding layer that enables the switch to detect and correct bit errors introduced over the cable between two interfaces. Because 25G transmission speeds can introduce a higher than acceptable bit error rate (BER) on a link, FEC is required or recommended for 25G, 4x25G, and 100G link speeds.

For the link to come up, the two interfaces on each end must use the same FEC setting.

There is a very small latency overhead required for FEC. For most applications, this small amount of latency is preferable to error packet retransmission latency.

There are two FEC types:

Cumulus Linux includes additional FEC options:

The Trident II switch does not support FEC.

The Tomahawk switch does not support RS FEC or auto-negotiation of FEC on 25G lanes that are broken out (Tomahawk pre-dates 802.3by). If you are using a 4x25G breakout DAC or AOC on a Tomahawk switch, you can configure either Base-R FEC or no FEC, and choose cables appropriate for that limitation (CA-25G-S, CA-25G-N or fiber). Tomahawk+, Tomahawk2, Trident3 and Maverick switches do not have this limitation.

For 25G DAC, 4x25G Breakouts DAC and 100G DAC cables, the IEEE 802.3by specification creates 3 classes:

The IEEE classification is based on various dB loss measurements and minimum achievable cable length. You can build longer and shorter cables if they comply to the dB loss and BER requirements.

If a cable is manufactured to CA-25G-S classification and FEC is not enabled, the BER might be unacceptable in a production network. It is important to set the FEC according to the cable class (or better) to have acceptable bit error rates. See Determining Cable Class below.

You can check bit errors using cl-netstat (RX_ERR column) or ethtool -S (HwIfInErrors counter) after a large amount of traffic has passed through the link. A non-zero value indicates bit errors. Expect error packets to be zero or extremely low compared to good packets. If a cable has an unacceptable rate of errors with FEC enabled, replace the cable.

For 25G, 4x25G Breakout, and 100G Fiber modules and AOCs, there is no classification of 25G cable types for dB loss, BER or length. FEC is recommended but might not be required if the BER is low enough.

Determine Cable Class of 100G and 25G DACs

You can determine the cable class for 100G and 25G DACs from the Extended Specification Compliance Code field (SFP28: 0Ah, byte 35, QSFP28: Page 0, byte 192) in the cable EEPROM programming.

For 100G DACs, most manufacturers use the 0x0Bh 100GBASE-CR4 or 25GBASE-CR CA-L value (the 100G DAC specification predates the IEEE 802.3by 25G DAC specification). RS FEC is the expected setting for 100G DAC but might not be required with shorter or better cables.

A manufacturer’s EEPROM setting might not match the dB loss on a cable or the actual bit error rates that a particular cable introduces. Use the designation as a guide, but set FEC according to the bit error rate tolerance in the design criteria for the network. For most applications, the highest mutual FEC ability of both end devices is the best choice.

You can determine for which grade the manufacturer has designated the cable as follows.

For the SFP28 DAC, run the following command:

cumulus@switch:~$ sudo ethtool -m swp35 hex on | grep 0020 | awk '{ print $6}'
0c

The values at location 0x0024 are:

For the QSFP28 DAC, run the following command:

cumulus@switch:~$ sudo ethtool -m swp51s0 hex on | grep 00c0 | awk '{print $2}'
0b

The values at 0x00c0 are:

In each example below, the Compliance field is derived using the method described above and is not visible in the ethool -m output.

3meter cable that does not require FEC 
(CA-N)  
Cost: More expensive  
Cable size: 26AWG (Note that AWG does not necessarily correspond to overall dB loss or BER performance)  
Compliance Code: 25GBASE-CR CA-N

3meter cable that requires Base-R FEC 
(CA-S)  
Cost: Less expensive  
Cable size: 26AWG  
Compliance Code: 25GBASE-CR CA-S

When in doubt, consult the manufacturer directly to determine the cable classification.

Spectrum ASIC FEC Behavior

The firmware in a Spectrum ASIC applies an FEC configuration to 25G and 100G cables based on the cable type and whether the peer switch also has a Spectrum ASIC.

When the link is between two switches with Spectrum ASICs:

Cable Type
FEC Mode
25G optical cablesBase-R/FC-FEC
25G 1,2 meters: CA-N, loss <13dbBase-R/FC-FEC
25G 2.5,3 meters: CA-S, loss <16dbBase-R/FC-FEC
25G 2.5,3,4,5 meters: CA-L, loss > 16dbRS-FEC
100G DAC or opticalRS-FEC

When linking to a non-Spectrum peer, the firmware lets the peer decide. The Spectrum ASIC supports RS-FEC (for both 100G and 25G), Base-R/FC-FEC (25G only), or no-FEC (for both 100G and 25G).

Cable Type
FEC Mode
25G pptical cablesLet peer decide
25G 1,2 meters: CA-N, loss <13dbLet peer decide
25G 2.5,3 meters: CA-S, loss <16dbLet peer decide
25G 2.5,3,4,5 meters: CA-L, loss > 16dbLet peer decide
100GLet peer decide: RS-FEC or No FEC

How Does Cumulus Linux use FEC?

This depends upon the make of the switch you are using.

A Spectrum switch enables FEC automatically when it powers up; that is, the setting is fec auto. The port firmware tests and determines the correct FEC mode to bring the link up with the neighbor. It is possible to get a link up to a Spectrum switch without enabling FEC on the remote device as the switch eventually finds a working combination to the neighbor without FEC.

On a Broadcom switch, Cumulus Linux does not enable FEC by default; that is, the setting is fec off. Cumulus Networks recommends you configure FEC explicitly to match the configured FEC on the link neighbor. On 100G DACs, you can configure link-autoneg so that the port attempts to negotiate FEC settings with the remote peer.

The following sections describe how to show the current FEC mode, and to enable and disable FEC.

Show the Current FEC Mode

Cumulus Linux returns different output for the ethtool --show-fec command, depending upon whether you are using a Broadcom or Mellanox Spectrum switch.

On a Broadcom switch, the --show-fec output tells you exactly what you configured, even if the link is down due to a FEC mismatch with the neighbor.

On a Spectrum switch, the --show-fec output tells you the current active state of FEC only if the link is up; that is, if the FEC modes matches that of the neighbor. If the link is not up, the value displays None, which is not valid.

To show the FEC mode currently enabled on a given switch port, run the ethtool --show-fec <interface> command.

cumulus@switch:~$ sudo ethtool --show-fec swp23
FEC parameters for swp23:
Configured FEC encodings: Auto
Active FEC encoding: Off

Enable or Disable FEC

To enable Reed Solomon (RS) FEC on a link:

NCLU Commands

Run the net add interface <interface> link fec rs command. For example:

cumulus@switch:~$ sudo net add interface swp23 link fec rs
cumulus@switch:~$ sudo net pending
cumulus@switch:~$ sudo net commit
Linux Commands
  1. Edit the /etc/network/interfaces file. The following example enables RS FEC for the swp1 interface (link-fec rs):
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
     link-autoneg off
     link-speed 100000
     link-fec rs
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

Run the ethtool --set-fec <interface> encoding RS command. For example:

cumulus@switch:~$ sudo ethtool --set-fec swp1 encoding RS

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To enable Base-R/FireCode FEC on a link:

NCLU Commands

Run the net add interface <interface> link fec baser command. For example:

cumulus@switch:~$ sudo net add interface swp23 link fec baser
cumulus@switch:~$ sudo net pending
cumulus@switch:~$ sudo net commit
Linux Commands
  1. Edit the /etc/network/interfaces file. The following example enables Base-R FEC for the swp1 interface (link-fec baser):
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
     link-autoneg off
     link-speed 100000
     link-fec baser
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

Run the ethtool --set-fec <interface> encoding baser command. For example:

cumulus@switch:~$ sudo ethtool --set-fec swp1 encoding BaseR

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To enable FEC with Auto-negotiation:

FEC with auto-negotiation is supported on DACs only.

NCLU Commands

Run the net add interface <interface> link autoneg on command. The following example command enables FEC with auto-negotiation on the swp12 interface:

cumulus@switch:~$ sudo net add interface swp12 link autoneg on
cumulus@switch:~$ sudo net pending
cumulus@switch:~$ sudo net commit
Linux Commands
  1. Edit the /etc/network/interfaces file and set auto-negotiation to on. For example:
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
    link-autoneg on
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

You can use ethtool to enable FEC with auto-negotiation. For example:

ethtool -s swp1 speed 10000 duplex full autoneg on

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To show the FEC and auto-negotiation settings for an interface, run the following command:

cumulus@switch:~$ sudo ethtool swp12 | egrep 'FEC|auto'
Supports auto-negotiation: Yes
Supported FEC modes: RS
Advertised auto-negotiation: Yes
Advertised FEC modes: RS
Link partner advertised auto-negotiation: Yes
Link partner advertised FEC modes: Not reported

To disable FEC on a link:

NCLU Commands

Run the net add interface <interface> link fec off command. For example:

cumulus@switch:~$ sudo net add interface swp23 link fec off
cumulus@switch:~$ sudo net pending
cumulus@switch:~$ sudo net commit
Linux Commands
  1. Edit the /etc/network/interfaces file. The following example disables Base-R FEC for the swp1 interface (link-fec baser):
cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp23
iface swp23
link-fec off
  1. Run the ifreload -a command to load the updated configuration:
cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

Run the ethtool --set-fec <interface> encoding off command. For example:

cumulus@switch:~$ sudo ethtool --set-fec swp23 encoding off

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

Interface Configuration Recommendations for Broadcom Platforms

The recommended configuration for each type of interface is described in the following table. These are the link settings that are applied to the port hardware when auto-negotiation is enabled on a Broadcom-based switch. If further troubleshooting is required to bring a link up, use the table below as a guide to set the link parameters.

Except as noted below, the settings for both sides of the link are expected to be the same.

Spectrum switches automatically configure these settings following a predefined list of parameter settings until the link comes up.

SpeedAuto-negotiationFEC SettingManual Configuration ExamplesNotes
100BASE-T (RJ-45 SFP adapter)OffN/ANCLU commands
$ net add interface swp1 link speed 100
$ net add interface swp1 link autoneg off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 100
The module has two sets of electronics: the port side, which communicates with the switch ASIC and the RJ-45 adapter side.

Auto-negotiation is always used on the RJ-45 adapter side of the link by the PHY built into the module. This is independent of the switch setting. Set auto-negotiation to off.

Auto-negotiation must be enabled on the server side in this scenario.
100BASE-T on a 1G fixed copper portOnN/ANCLU commands
$ net add interface swp1 link speed 100
$net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   ink-autoneg on
   link-speed 100
10M or 100M speeds are possible with auto-negotiation off on both sides.

Testing on an Edgecore AS4610-54P showed the ASIC reporting auto-negotiation as on.

Power over Ethernet might require auto-negotiation to be on.
1000BASE-T (RJ-45 SFP adapter)OffN/ANCLU commands
$ net add interface swp1 link speed 1000
$ net add interface swp1 link autoneg off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 1000
The module has two sets of electronics: the port side, which communicates with the switch ASIC and the RJ-45 side.

Auto-negotiation is always used on the RJ-45 side of the link by the PHY built into the module. This is independent of the switch setting. Set auto-negotiation to off.

Auto-negotiation must be enabled on the server side.
1000BASE-T on a 1G fixed copper portOnN/ANCLU commands
$ net add interface swp1 link speed 1000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 1000
1000BASE-T on a 10G fixed copper portOnN/ANCLU commands
$ net add interface swp1 link speed 1000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 1000
1000BASE-SX 1000BASE-LX (1G Fiber)Recommended OnN/ANCLU commands
$ net add interface swp1 link speed 1000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 1000
Without auto-negotiation, the link stays up when there is a single fiber break.

See the limitation discussed in 10G and 1G SFPs Inserted in a 25G Port below.
10GBASE-T (RJ-45 SFP Module)OffN/ANCLU commands
$ net add interface swp1 link speed 10000
$ net add interface swp1 link autoneg off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 10000
The module has two sets of electronics - the port side, which communicates to the switch ASIC and the RJ-45 side.

Auto-negotiation is always used on the RJ-45 side of the link by the PHY built into the module. This is independent of the switch setting. Set link-autoneg to off.

Auto-negotiation needs to be enabled on the server side.
10GBASE-T fixed copper portOnN/ANCLU commands
$ net add interface swp1 link speed 10000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 10000
10GBASE-CR
10GBASE-LR
10GBASE-SR
10G AOC
OffN/ANCLU commands
$ net add interface swp1 link speed 10000
$ net add interface swp1 link autoneg off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 10000
40GBASE-CR4Recommended OnDisableNCLU commands
$ net add interface swp1 link speed 40000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 40000
40G standards mandate auto-negotiation be enabled for DAC connections.
40GBASE-SR4
40GBASE-LR4
40G AOC
OffDisableNCLU commands
$ net add interface swp1 link speed 40000
$ net add interface swp1 link autoneg off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 40000
100GBASE-CR4Onauto-negotiatedNCLU commands
$ net add interface swp1 link speed 100000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 100000
100GBASE-SR4
100G AOC
OffRSNCLU commands
$ net add interface swp1 link speed 100000
$ net add interface swp1 link autoneg off
$ net add interface swp1 link fec rs
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 100000
   link-fec rs
100GBASE-LR4OffNoneNCLU commands
$ net add interface swp1 link speed 100000
$ net add interface swp1 link autoneg off
$ net add interface swp1 link fec off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 100000
   link-fec off
25GBASE-CROnauto-negotiatedNCLU commands
$ net add interface swp1 link speed 25000
$ net add interface swp1 link autoneg on
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg on
   link-speed 25000
Tomahawk predates 802.3by. It does not support RS FEC or auto-negotiation of RS FEC on a 25G port or subport. It does support Base-R FEC.
25GBASE-SROffRSNCLU commands
$ net add interface swp1 link speed 25000
$ net add interface swp1 link autoneg off
$ net add interface swp1 link fec baser
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
   link-speed 25000
   link-fec baser
Tomahawk predates 802.3by. It does not support RS FEC on a 25G port or subport. It does support Base-R FEC.
25GBASE-LROffNoneNCLU commands
$ net add interface swp1 link speed 25000
$ net add interface swp1 link autoneg off
$ net add interface swp1 link fec off
Configuration in /etc/network/interfaces
auto swp1
iface swp1
   link-autoneg off
    link-speed 25000
   link-fec off

Default Policies for Interface Settings

Instead of configuring settings for each individual interface, you can specify a policy for all interfaces on a switch or tailor custom settings for each interface. Create a file in /etc/network/ifupdown2/policy.d/ and populate the settings accordingly. The following example shows a file called address.json.

cumulus@switch:~$ cat /etc/network/ifupdown2/policy.d/address.json
{
    "ethtool": {
        "defaults": {
            "link-duplex": "full"
        },
        "iface_defaults": {
            "swp1": {
                "link-autoneg": "on",
                "link-speed": "1000"
          },
            "swp16": {
                "link-autoneg": "off",
                "link-speed": "10000"
            },
            "swp50": {
                "link-autoneg": "off",
                "link-speed": "100000",
                "link-fec": "rs"
            }
        }
    },
    "address": {
        "defaults": { "mtu": "9000" },
        "iface_defaults": {
            "eth0": {"mtu": "1500"}
        }
    }
}

Setting the default MTU also applies to the management interface. Be sure to add the iface_defaults to override the MTU for eth0, to remain at 1500 for Broadcom switches or 9238 for Mellanox switches.

Breakout Ports

Cumulus Linux provides the ability to:

  • For switches with ports that support 100G speeds, you can break out any 100G port into a variety of options: four 10G ports, four 25G ports, two 40G ports or two 50G ports. You cannot have more than 128 total logical ports on a Broadcom switch.
  • You can only use NCLU to configure a 4x25G breakout port. To configure other breakout ports, use Linux commands.
  • You cannot use NCLU to break out the uplink ports.

The Mellanox SN2700, SN2700B, SN2410, and SN2410B switches all have a limit of 64 logical ports in total. However, if you want to break out to 4x25G or 4x10G, you must configure the logical ports as follows:

  • You can only break out odd-numbered ports into 4 logical ports.
  • You must disable the next even-numbered port. For example, if you break out port 11 into 4 logical ports, you must disable port 12.

These restrictions do not apply to a 2x50G breakout configuration.

To configure a breakout port:

NCLU Commands - 4x25G breakout ports only

Run the following commands to configure the port to break out and set the link speed. The following example command breaks out swp3 into four 25G ports:

cumulus@switch:~$ net add interface swp3 breakout 4x25G
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

On Spectrum switches, you need to disable the next port. The following example command disables swp4.

cumulus@switch:~$ net add interface swp4 breakout disabled
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands break out the 100G interfaces to 4x25G interfaces in the /etc/cumulus/ports.conf file, restart the switchd process to reconfigure the ports and create four interfaces in the /etc/network/interfaces file named as follows:

cumulus@switch:~$ cat /etc/network/interfaces
...
auto swp3s0
iface swp3s0

auto swp3s1
iface swp3s1

auto swp3s2
iface swp3s2

auto swp3s3
iface swp3s3
...

The breakout port configuration is stored in the /etc/cumulus/ports.conf file.

When you commit your change, switchd restarts to apply the changes. The restart interrupts network services.

Linux Commands
  1. Edit the /etc/cumulus/ports.conf file to configure the port breakout. See the examples below.
  2. Configure the breakout ports in the /etc/network/interfaces file. See the example below.
  3. Restart switchd.

The /etc/cumulus/ports.conf file varies across different hardware platforms. Check the current list of supported platforms on (the hardware compatibility list.

The following example shows a snippet from the /etc/cumulus/ports.conf file on a Dell Z9264F-ON switch (with a Tomahawk2 ASIC) where swp1 and swp63 are broken out into four 10G ports:

cumulus@switch:~$ sudo cat /etc/cumulus/ports.conf
# ports.conf --
#
#   configure port speed, aggregation, and subdivision.
#
# The Dell Z9264F has:
#      64 QSFP28 ports numbered 1-64
#         These ports are configurable as 100G, 50G, 40G, or split into
#         2x50G, 4x25G, or 4x10G ports.
#
# NOTE:  You must restart switchd for any changes to take effect.
# Only "odd-numbered " port can be split into 4 interfaces and if an odd-numbered
# port is split in a 4X configuration, the port adjacent to it (even-numbered port)
# has to be set to "disabled " in this file. When splitting a port into two
# interfaces, like 2x50G, it is NOT required that the adjacent port be
# disabled. For example, when splitting port 11 into 4 10G interfaces, port
# 12 must be configured as "disabled" like this:
#
#   11=4x10G
#   12=disabled

# QSFP28 ports
#
# <port label> = [100G|50G|40G|2x50G|4x25G|4x10G|disabled]

1=4x10G
2=disabled
3=100G
4=100G
5=100G
6=100G
7=100G
8=100G
9=100G
10=100G
11=100G
12=100G
13=100G
14=100G
15=100G
16=100G
17=100G
18=100G
19=100G
20=100G
21=100G
22=100G
23=100G
24=100G
25=100G
26=100G
27=100G
28=100G
29=100G
30=100G
31=100G
32=100G
33=100G
34=100G
35=100G
36=100G
37=100G
38=100G
39=100G
40=100G
41=100G
42=100G
43=100G
44=100G
45=100G
46=100G
47=100G
48=100G
49=100G
50=100G
51=100G
52=100G
53=100G
54=100G
55=100G
56=100G
57=100G
58=100G
59=100G
60=100G
61=100G
62=100G
63=4x10G
64=disabled

The following example shows the swp26 breakout ports (swp26s0, swp26s1, swp26s2, and swp26s3) in the /etc/network/interfaces file.

cumulus@switch:~$ sudo cat /etc/network/interfaces
...
auto swp26s0
iface swp26s0

auto swp26s1
iface swp26s1

auto swp26s2
iface swp26s2

auto swp26s3
iface swp26s3
...

Refer to this article for an example of how to configure breakout cables for the Mellanox Spectrum SN2700 switch.

Break out a 100G Port to Four 10G Ports

If you want to support 10G speed modules or cables on 100G ports you must set up the port in 10G mode first by configuring breakout ports on the 100G ports using the following commands:

NCLU Commands
cumulus@switch:~$ net add interface swp25 breakout 4x10G
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands
  1. Edit the /etc/cumulus/ports.conf file to configure the port breakout.
cumulus@switch:~$ sudo nano /etc/cumulus/ports.conf
...

25s0=10G
25s1=10G
25s2=10G
25s3=10G
...
  1. Configure the breakout ports in the /etc/network/interfaces file.
cumulus@switch:~$ sudo nano /etc/network/interfaces
...

auto swp25s0
iface swp25s0

auto swp25s1
iface swp25s1

auto swp25s2
iface swp25s2

auto swp25s3
iface swp25s3
...
  1. Restart switchd.

Remove a Breakout Port

To remove a breakout port:

NCLU Commands
  1. Run the net del interface <interface> command. For example:
cumulus@switch:~$ net del interface swp3s0
cumulus@switch:~$ net del interface swp3s1
cumulus@switch:~$ net del interface swp3s2
cumulus@switch:~$ net del interface swp3s3
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
  1. Manually edit the /etc/cumulus/ports.conf file to configure the interface for the original speed. For example:
cumulus@switch:~$ sudo nano /etc/cumulus/ports.conf
...

2=100G
3=100G
4=100G
...
  1. Restart switchd.
Linux Commands
  1. Edit the /etc/cumulus/ports.conf file to configure the interface for the original speed.
cumulus@switch:~$ sudo nano /etc/cumulus/ports.conf
...

2=100G
3=100G
4=100G
...
  1. Restart switchd.

Combine Four 10G Ports into One 40G Port

You can gang (combine) four 10G ports into one 40G port for use with a breakout cable, provided you follow these requirements:

The /etc/cumulus/ports.conf file varies across different hardware platforms. Check the current list of supported platforms on the hardware compatibility list.

NCLU Commands

To gang swp1 through swp4 into a 40G port, run the following commands:

cumulus@switch:~$ net add int swp1-4 breakout /4
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration snippet in the /etc/cumulus/ports.conf file:

# SFP+ ports#
# <port label 1-48> = [10G|40G/4]
1=40G/4
2=40G/4
3=40G/4
4=40G/4
5=10G
Linux Commands

To gang swp1 through swp4 into a 40G port, edit the /etc/cumulus/ports.conf file as shown below, then restart switchd.

# SFP+ ports#
# <port label 1-48> = [10G|40G/4]
1=40G/4
2=40G/4
3=40G/4
4=40G/4
5=10G

Logical Switch Port Limitations

100G and 40G switches can support a certain number of logical ports, depending on the manufacturer; these include:

Before you configure any logical/unganged ports on a switch, check the limitations listed in /etc/cumulus/ports.conf; this file is specific to each manufacturer.

The following example shows the logical port limitation provided in the Dell Z9254F-ON ports.conf file. The maximum number of ports for this switch is 128.

# ports.conf --
#
#   configure port speed, aggregation, and subdivision.
#
# The Dell Z9264F has:
#      64 QSFP28 ports numbered 1-64
#         These ports are configurable as 100G, 50G, 40G, or split into
#         2x50G, 4x25G, or 4x10G ports.
#
# NOTE:  You must restart switchd for any changes to take effect.
# Only "odd-numbered " port can be split into 4 interfaces and if an odd-numbered
# port is split in a 4X configuration, the port adjacent to it (even-numbered port)
# has to be set to "disabled " in this file. When splitting a port into two
# interfaces, like 2x50G, it is NOT required that the adjacent port be
# disabled. For example, when splitting port 11 into 4 10G interfaces, port
# 12 must be configured as "disabled" like this:
#
#   11=4x10G
#   12=disabled

# QSFP28 ports
#
# <port label> = [100G|50G|40G|2x50G|4x25G|4x10G|disabled]

Mellanox SN2700 and SN2700B switches have a limit of 64 logical ports in total. However, the logical ports must be configured in a specific way. See the note above.

Verification and Troubleshooting Commands

Statistics

To show high-level interface statistics, run the net show interface command:

cumulus@switch:~$ net show interface swp1

    Name    MAC                Speed      MTU  Mode
--  ------  -----------------  -------  -----  ---------
UP  swp1    44:38:39:00:00:04  1G        1500  Access/L2

Vlans in disabled State
-------------------------
br0

Counters      TX    RX
----------  ----  ----
errors         0     0
unicast        0     0
broadcast      0     0
multicast      0     0

LLDP
------  ----  ---------------------------
swp1    ====  44:38:39:00:00:03(server01)

To show low-level interface statistics, run the following ethtool command:

cumulus@switch:~$ sudo ethtool -S swp1
NIC statistics:
      HwIfInOctets: 21870
      HwIfInUcastPkts: 0
      HwIfInBcastPkts: 0
      HwIfInMcastPkts: 243
      HwIfOutOctets: 1148217
      HwIfOutUcastPkts: 0
      HwIfOutMcastPkts: 11353
      HwIfOutBcastPkts: 0
      HwIfInDiscards: 0
      HwIfInL3Drops: 0
      HwIfInBufferDrops: 0
      HwIfInAclDrops: 0
      HwIfInBlackholeDrops: 0
      HwIfInDot3LengthErrors: 0
      HwIfInErrors: 0
      SoftInErrors: 0
      HwIfOutErrors: 0
      HwIfOutQDrops: 0
      HwIfOutNonQDrops: 0
      SoftOutErrors: 0
      SoftOutDrops: 0
      SoftOutTxFifoFull: 0
      HwIfOutQLen: 0

Query SFP Port Information

To verify SFP settings, run the ethtool -m command. The following example shows the vendor, type and power output for the swp4 interface.

cumulus@switch:~$ sudo ethtool -m swp4 | egrep 'Vendor|type|power\s+:'
        Transceiver type                          : 10G Ethernet: 10G Base-LR
        Vendor name                               : FINISAR CORP.
        Vendor OUI                                : 00:90:65
        Vendor PN                                 : FTLX2071D327
        Vendor rev                                : A
        Vendor SN                                 : UY30DTX
        Laser output power                        : 0.5230 mW / -2.81 dBm
        Receiver signal average optical power     : 0.7285 mW / -1.38 dBm

Caveats and Errata

Port Speed and the ifreload -a Command

When configuring port speed or break outs in the /etc/cumulus/ports.conf file, you need to run the ifreload -a command to reload the configuration after restarting switchd in the following cases:

Port Speed Configuration

If you change the port speed in the /etc/cumulus/ports.conf file but the speed is also configured for that port in the /etc/network/interfaces file, after you edit the /etc/cumulus/ports.conf file and restart switchd, you must also run the ifreload -a command so that the /etc/network/interfaces file is also updated with your change.

10G and 1G SFPs Inserted in a 25G Port

For 10G and 1G SFPs inserted in a 25G port on a Broadcom platform, you must configure the four ports in the same core to be 10G. Each set of four 25G ports are controlled by a single core; therefore, each core must run at the same clock speed. The four ports must be in sequential order; for example, swp1, swp2, swp3, and swp4.

  1. Edit the /etc/cumulus/ports.conf file and configure the four ports to be 10G. 1G SFPs are clocked at 10G speeds; therefore, for 1G SFPs, the /etc/cumulus/ports.conf file entry must also specify 10G. Currently you cannot use NCLU commands for this step.
...
# SFP28 ports
#
# <port label 1-48> = [25G|10G|100G/4|40G/4]
1=25G
2=25G
3=25G
4=25G
5=10G
6=10G
7=10G
8=10G
9=25G
...
  1. Restart switchd.
  2. If you want to set the speed of any SFPs to 1G, set the port speed to 1000 Mbps using NCLU commands; this is not necessary for 10G SFPs. You don’t need to set the port speed to 1G for all four ports. For example, if you intend only for swp5 and swp6 to use 1G SFPs, do the following:
cumulus@switch:~$ net add interface swp5-swp6 link speed 1000
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

100G switch ASICs do not support 1000Base-X auto-negotiation (Clause 37), which is recommended for 1G fiber optical modules. As a result, single fiber breaks cannot be detected when using 1G optical modules on these switches.

The auto-negotiation setting must be the same on both sides of the connection. If using 1G fiber modules in 25G SFP28 ports, ensure auto-negotiation is disabled on the link partner interface as well.

Timeout Error on Quanta LY8 and LY9 Switches

On Quanta T5048-LY8 and T3048-LY9 switches, an Operation timed out error occurs when you remove and reinsert a QSFP module.

You cannot remove the QSFPx2 module while the switch is powered on; it is not hot-swappable. However, if an Operation timed out error occurs, restart switchd to bring the link up. Be aware that this disrupts your network.

On the T3048-LY9, run the following commands:

cumulus@switch:~$ sudo echo 0 > qsfpd_power_enable/value
cumulus@switch:~$ sudo rmmod quanta_ly9_rangeley_platform
cumulus@switch:~$ sudo modprobe quanta_ly9_rangeley_platform
cumulus@switch:~$ sudo systemctl restart switchd.service

On the T5048-LY8, run the following commands:

cumulus@switch:~$ sudo echo 0 > qsfpd_power_enable/value
cumulus@switch:~$ sudo systemctl restart switchd.service

swp33 and swp34 Disabled on Some Switches

The front SFP+ ports (swp33 and swp34) are disabled in Cumulus Linux on the following switches:

These ports appear as disabled in the /etc/cumulus/ports.conf file.

200G Interfaces on the Dell S5248F Switch

On the Dell S5248F switch, the 2x200G QSFP-DD interfaces labeled 49/50 and 51/52 are not supported natively at 200G speeds. The interfaces are supported with 100G cables; however, you can only use one 100G from each QSFP-DD port. The upper QSFP-DD port is named swp49 and the lower QSFP-DD port is named swp52.

QSFP+ Ports on the Dell S5232F Switch

Cumulus Linux does not support the 2x10G QSFP+ ports on the Dell S5232F switch.

QSFP+ Ports on the Dell S4148T Switch

On the Dell S4148T switch, the two QSFP+ ports are set to disabled by default and the four QSFP28 ports are configured for 100G. The following example shows the default settings in the /etc/cumulus/ports.conf file for this switch:

cumulus@switch:~$ sudo cat /etc/cumulus/ports.conf
...
# QSFP+ ports
#
# <port label 27-28> = [4x10G|40G]
27=disabled
28=disabled
# QSFP28 ports
#
# <port label 25-26, 29-30> = [4x10G|4x25G|2x50G|40G|50G|100G]
25=100G
26=100G
29=100G
30=100G

To enable the two QSFP+ ports, you must configure all four QSFP28 ports for either 40G or 4x10G. You cannot use either of the QSFP+ ports if any of the QSFP28 ports are configured for 100G.

The following example shows the /etc/cumulus/ports.conf file with all four QSFP28 ports configured for 40G and both QSFP+ ports enabled:

cumulus@switch:~$ sudo cat /etc/cumulus/ports.conf
...
# QSFP+ ports
#
# <port label 27-28> = [4x10G|40G]
27=40G
28=40G
# QSFP28 ports
#
# <port label 25-26, 29-30> = [4x10G|4x25G|2x50G|40G|50G|100G]
25=40G
26=40G
29=40G
30=40G

To disable the QSFP+ ports, you must set the ports to disabled. Do not comment out the lines as this prevents switchd from restarting.

1000BASE-T Modules Not Supported on Certain Edgecore and Cumulus Express Switches

1000BASE-T modules are not supported on the following switches:

After rebooting the Melllanox SN2100 switch, eth0 always has a speed of 100Mb/s. If you bring the interface down and then back up again, the interface negotiates 1000Mb. This only occurs the first time the interface comes up.

To work around this issue, add the following commands to the /etc/rc.local file to flap the interface automatically when the switch boots:

modprobe -r igb
sleep 20
modprobe igb

On the EdgeCore AS7326-56X switch, all four switch ports in each port group must be set to the same link speed; otherwise, the links do not come up. These ports are set to 25G by default, but can also be set to 10G. The port groups on this switch are as follows, where each row is a port group:

For example, if you configure port 19 for 10G, you must also configure ports 16, 17 and 21 for 10G.

Additionally, you can gang each port group together as a 100G or 40G port. When ganged together, one port (based on the arrangement of the ports) is designated as the gang leader. This port’s number is used to configure the ganged ports and is marked with an asterisk (*) above.

The EdgeCore AS7326-56X is a 48x25G + 8x100G + 2x10G switch. The dedicated 10G ports are not currently supported in Cumulus Linux. However, you can configure all other ports to run at 10G speeds.

The Lenovo NE2572O switch has external retimers on swp1 through swp8. Currently, these ports only support a speed of 25G.

The following switches that use Serial over LAN technology (SOL) do not support eth0 speed or auto-negotiation changes:

ethtool Shows Incorrect Port Speed on 100G Spectrum Switches

On a Spectrum switch, after you set the interface speed to 40G in the ports.conf file, ethtool still shows the speed as 100G. This is a known issue where ethtool does not update after restarting switchd and continues to display the outdated port speed.

To correctly set the port speed, run the following commands.

NCLU Commands

Run the net add interface <interface> link speed command. The following example command sets the port speed to 40G:

cumulus@switch:~$ net add interface swp1 link speed 40000
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Run the ethtool -s <interface> speed command. The following example command sets the port speed to 40G:

cumulus@switch:~$ sudo ethtool -s swp1 speed 40000

Delay in Reporting Interface as Operational Down

When you remove two transceivers simultaneously from a switch, both interfaces show the carrier down status immediately. However, it takes one second for the second interface to show the operational down status. In addition, the services on this interface also take an extra second to come down.

Mellanox Spectrum-2 and Tomahawk-based Switches Support Different FEC Modes

The Mellanox Spectrum-2 (25G) switch only supports RS FEC. The Tomahawk-based switch only supports BASE-R FEC. These two switches do not share compatible FEC modes and do not interoperate reliably.

Maverick Switches with Modules that Don’t Support Auto-negotiation

On a Maverick switch, if auto-negotiation is configured on a 10G interface and the installed module does not support auto-negotiation (for example, 10G DAC, 10G Optical, 1G RJ45 SFP), the link breaks. To work around this issue, disable auto-negotiation on interfaces where it is not supported.

ifplugd

ifplugd is an Ethernet link-state monitoring daemon that executes user-specified scripts to configure an Ethernet device when a cable is plugged in, or automatically unconfigure an Ethernet device when a cable is removed. Follow the steps below to install and configure the ifplugd daemon.

Install ifplugd

You can install this package even if the switch is not connected to the internet, as it is contained in the cumulus-local-apt-archive repository that is embedded in the Cumulus Linux disk image.

To install ifplugd:

  1. Update the switch before installing the daemon:
cumulus@switch:~$ sudo -E apt-get update
  1. Install the ifplugd package:
cumulus@switch:~$ sudo -E apt-get install ifplugd

Configure ifplugd

After you install ifplugd, you must edit two configuration files:

The example configuration below configures ifplugd to bring down all uplinks when the peer bond goes down in an MLAG environment.

  1. Open /etc/default/ifplugd in a text editor and configure the file as appropriate. Add the peerbond name before you save the file.
INTERFACES="peerbond"
HOTPLUG_INTERFACES=""
ARGS="-q -f -u0 -d1 -w -I"
SUSPEND_ACTION="stop"
  1. Open the /etc/ifplugd/action.d/ifupdown file in a text editor. Configure the script, then save the file.
#!/bin/sh
set -e
case "$2" in
up)
           clagrole=$(clagctl | grep "Our Priority" | awk '{print $8}')
           if [ "$clagrole" = "secondary" ]
           then
               #List all the interfaces below to bring up when clag peerbond comes up.
               for interface in swp1 bond1 bond3 bond4
               do
                   echo "bringing up : $interface"  
                   ip link set $interface up
               done
           fi
       ;;
down)
           clagrole=$(clagctl | grep "Our Priority" | awk '{print $8}')
           if [ "$clagrole" = "secondary" ]
           then
               #List all the interfaces below to bring down when clag peerbond goes down.
               for interface in swp1 bond1 bond3 bond4
               do
                   echo "bringing down : $interface"
                   ip link set $interface down
               done
           fi
       ;;
esac
  1. Restart the ifplugd daemon to implement the changes:
cumulus@switch:~$ sudo systemctl restart ifplugd.service

Caveats and Errata

The default shell for ifplugd is dash (/bin/sh) instead of bash, as it provides a faster and more nimble shell. However, dash contains fewer features than bash (for example, dash is unable to handle multiple uplinks).

Buffer and Queue Management

Hardware datapath configuration manages packet buffering, queueing and scheduling in hardware. To configre priority groups, and assign the scheduling alogorithm and weights, you edit the /etc/cumulus/datapath/traffic.conf.

The /usr/lib/python2.7/dist-packages/cumulus/__chip_config/[bcm|mlx]/datapath.conf assigns buffer space and egress queues. Cumulus Networks strongly recommends you work with a Cumulus support engineer to change buffer limits in the datapath.conf file.

Each packet is assigned to an ASIC Class of Service (CoS) value based on the priority value of the packet stored in the 802.1p (Class of Service) or DSCP (Differentiated Services Code Point) header field. The choice to schedule packets based on COS or DSCP is a configurable option in the /etc/cumulus/datapath/traffic.conf file.

Priority groups include:

The scheduler is configured to use a hybrid scheduling algorithm. It applies strict priority to control traffic queues and a weighted round robin selection from the remaining queues. Unicast packets and multicast packets with the same priority value are assigned to separate queues, which are assigned equal scheduling weights.

You can configure Quality of Service (QoS) for switches on the following platforms only:

  • Broadcom Tomahawk, Trident II, Trident II+ and Trident3
  • Mellanox Spectrum

Example Configuration File

The following example /etc/cumulus/datapath/traffic.conf datapath configuration file applies to 10G, 40G, and 100G switches on Broadcom Tomahawk, Trident II, Trident II+, or Trident3 and Mellanox Spectrum platforms only.

click to see traffic.conf file
cumulus@switch:~$ sudo cat /etc/cumulus/datapath/traffic.conf
#
# /etc/cumulus/datapath/traffic.conf
#

# packet header field used to determine the packet priority level
# fields include {802.1p, dscp}
traffic.packet_priority_source_set = [802.1p,dscp]

# remark packet priority value
# fields include {802.1p, none}
# remark packet priority value
# fields include {802.1p, dscp}
traffic.packet_priority_remark_set = [802.1p,dscp]

# packet priority remark values assigned from each internal cos value
# internal cos values {cos_0..cos_7}
# (internal cos 3 has been reserved for CPU-generated traffic)
#
# 802.1p values = {0..7}

traffic.cos_0.priority_remark.8021p = [1]
traffic.cos_1.priority_remark.8021p = [0]
traffic.cos_2.priority_remark.8021p = [3]
traffic.cos_3.priority_remark.8021p = [2]
traffic.cos_4.priority_remark.8021p = [4]
traffic.cos_5.priority_remark.8021p = [5]
traffic.cos_6.priority_remark.8021p = [7]
traffic.cos_7.priority_remark.8021p = [6]

# dscp values = {0..63}
traffic.cos_0.priority_remark.dscp = [1]
traffic.cos_1.priority_remark.dscp = [9]
traffic.cos_2.priority_remark.dscp = [17]
traffic.cos_3.priority_remark.dscp = [25]
traffic.cos_4.priority_remark.dscp = [33]
traffic.cos_5.priority_remark.dscp = [41]
traffic.cos_6.priority_remark.dscp = [49]
traffic.cos_7.priority_remark.dscp = [57]

# Per-port remark packet fields and mapping: applies to the designated set of ports.
remark.port_group_list = [remark_port_group]
remark.remark_port_group.packet_priority_remark_set = [802.1p,dscp]
remark.remark_port_group.port_set = swp1-swp4,swp6
remark.remark_port_group.cos_0.priority_remark.dscp = [2]
remark.remark_port_group.cos_1.priority_remark.dscp = [10]
remark.remark_port_group.cos_2.priority_remark.dscp = [18]
remark.remark_port_group.cos_3.priority_remark.dscp = [26]
remark.remark_port_group.cos_4.priority_remark.dscp = [34]
remark.remark_port_group.cos_5.priority_remark.dscp = [42]
remark.remark_port_group.cos_6.priority_remark.dscp = [50]
remark.remark_port_group.cos_7.priority_remark.dscp = [58]

# packet priority values assigned to each internal cos value
# internal cos values {cos_0..cos_7}
# (internal cos 3 has been reserved for CPU-generated traffic)
#
# 802.1p values = {0..7}
traffic.cos_0.priority_source.8021p = [0]
traffic.cos_1.priority_source.8021p = [1]
traffic.cos_2.priority_source.8021p = [2]
traffic.cos_3.priority_source.8021p = []
traffic.cos_4.priority_source.8021p = [3,4]
traffic.cos_5.priority_source.8021p = [5]
traffic.cos_6.priority_source.8021p = [6]
traffic.cos_7.priority_source.8021p = [7]

# dscp values = {0..63}
traffic.cos_0.priority_source.dscp = [0,1,2,3,4,5,6,7]
traffic.cos_1.priority_source.dscp = [8,9,10,11,12,13,14,15]
traffic.cos_2.priority_source.dscp = []
traffic.cos_3.priority_source.dscp = []
traffic.cos_4.priority_source.dscp = []
traffic.cos_5.priority_source.dscp = []
traffic.cos_6.priority_source.dscp = []
traffic.cos_7.priority_source.dscp = [56,57,58,59,60,61,62,63]

# Per-port source packet fields and mapping: applies to the designated set of ports.
source.port_group_list = [source_port_group]
source.source_port_group.packet_priority_source_set = [802.1p,dscp]
source.source_port_group.port_set = swp1-swp4,swp6
source.source_port_group.cos_0.priority_source.8021p = [7]
source.source_port_group.cos_1.priority_source.8021p = [6]
source.source_port_group.cos_2.priority_source.8021p = [5]
source.source_port_group.cos_3.priority_source.8021p = [4]
source.source_port_group.cos_4.priority_source.8021p = [3]
source.source_port_group.cos_5.priority_source.8021p = [2]
source.source_port_group.cos_6.priority_source.8021p = [1]
source.source_port_group.cos_7.priority_source.8021p = [0]

# priority groups
traffic.priority_group_list = [control, service, bulk]

# internal cos values assigned to each priority group
# each cos value should be assigned exactly once
# internal cos values {0..7}
priority_group.control.cos_list = [7]
priority_group.service.cos_list = [2]
priority_group.bulk.cos_list = [0,1,3,4,5,6]

# to configure priority flow control on a group of ports:
# -- assign cos value(s) to the cos list
# -- add or replace a port group names in the port group list
# -- for each port group in the list
#    -- populate the port set, e.g.
#       swp1-swp4,swp8,swp50s0-swp50s3
#    -- set a PFC buffer size in bytes for each port in the group
#    -- set the xoff byte limit (buffer limit that triggers PFC frame transmit to start)
#    -- set the xon byte delta (buffer limit that triggers PFC frame transmit to stop)
#    -- enable PFC frame transmit and/or PFC frame receive
# priority flow control
# pfc.port_group_list = [pfc_port_group]
# pfc.pfc_port_group.cos_list = []
# pfc.pfc_port_group.port_set = swp1-swp4,swp6
# pfc.pfc_port_group.port_buffer_bytes = 25000
# pfc.pfc_port_group.xoff_size = 10000
# pfc.pfc_port_group.xon_delta = 2000
# pfc.pfc_port_group.tx_enable = true
# pfc.pfc_port_group.rx_enable = true

# to configure pause on a group of ports:
# -- add or replace port group names in the port group list
# -- for each port group in the list
#    -- populate the port set, e.g.
#       swp1-swp4,swp8,swp50s0-swp50s3
#    -- set a pause buffer size in bytes for each port in the group
#    -- set the xoff byte limit (buffer limit that triggers pause frames transmit to start)
#    -- set the xon byte delta (buffer limit that triggers pause frames transmit to stop)

# link pause
# link_pause.port_group_list = [pause_port_group]
# link_pause.pause_port_group.port_set = swp1-swp4,swp6
# link_pause.pause_port_group.port_buffer_bytes = 25000
# link_pause.pause_port_group.xoff_size = 10000
# link_pause.pause_port_group.xon_delta = 2000
# link_pause.pause_port_group.rx_enable = true
# link_pause.pause_port_group.tx_enable = true

# scheduling algorithm: algorithm values = {dwrr}
scheduling.algorithm = dwrr

# traffic group scheduling weight
# weight values = {0..127}
# '0' indicates strict priority
priority_group.control.weight = 0
priority_group.service.weight = 32
priority_group.bulk.weight = 16

# To turn on/off Denial of service (DOS) prevention checks
dos_enable = false

# Cut-through is disabled by default on all chips with the exception of
# Spectrum. On Spectrum cut-through cannot be disabled.
#cut_through_enable = false

# Enable resilient hashing
#resilient_hash_enable = FALSE

# Resilient hashing flowset entries per ECMP group
# Valid values - 64, 128, 256, 512, 1024
#resilient_hash_entries_ecmp = 128

# Enable symmetric hashing
#symmetric_hash_enable = TRUE

# Set sflow/sample ingress cpu packet rate and burst in packets/sec
# Values: {0..16384}
#sflow.rate = 16384  
#sflow.burst = 16384

#Specify the maximum number of paths per route entry.
#  Maximum paths supported is 200.
#  Default value 0 takes the number of physical ports as the max path size.
#ecmp_max_paths = 0

#Specify the hash seed for Equal cost multipath entries
# Default value 0
# Value Rang: {0..4294967295}
#ecmp_hash_seed = 42

# Specify the forwarding table resource allocation profile, applicable
# only on platforms that support universal forwarding resources.
#
# /usr/cumulus/sbin/cl-rsource-query reports the allocated table sizes
# based on the profile setting.
#
#   Values: one of {'default', 'l2-heavy', 'v4-lpm-heavy', 'v6-lpm-heavy'}
#   Default value: 'default'
#   Note: some devices may support more modes, please consult user
#         guide for more details
#
#forwarding_table.profile = default

On switches with Spectrum ASICs, you must enable packet priority remark on the ingress port. A packet received on a remark-enabled port is remarked according to the priority mapping configured on the egress port. If you configure packet priority remark the same way on every port, the default configuration example above is correct. However, per-port customized configurations require two port groups: one for the ingress ports and one for the egress ports, as below:

remark.port_group_list = [ingress_remark_group, egress_remark_group]
remark.ingress_remark_group.packet_priority_remark_set = [dscp]
remark.remark_port_group.port_set = swp1-swp4,swp6
remark.egress_remark_group.port_set = swp10-swp20
remark.egress_remark_group.cos_0.priority_remark.dscp = [2]
remark.egress_remark_group.cos_1.priority_remark.dscp = [10]
remark.egress_remark_group.cos_2.priority_remark.dscp = [18]
remark.egress_remark_group.cos_3.priority_remark.dscp = [26]
remark.egress_remark_group.cos_4.priority_remark.dscp = [34]
remark.egress_remark_group.cos_5.priority_remark.dscp = [42]
remark.egress_remark_group.cos_6.priority_remark.dscp = [50]
remark.egress_remark_group.cos_7.priority_remark.dscp = [58]

On Broadcom switches, if you modify the configuration in the /etc/cumulus/datapath/traffic.conf file, you must restart switchd for the changes to take effect; run the cumulus@switch:~$ sudo systemctl restart switchd.service command.

On Mellanox switches with the Spectrum ASIC, the following options in the /etc/cumulus/datapath/traffic.conf file do not require you to restart switchd. However, you must run the echo 1 > /cumulus/switchd/config/traffic/reload command after you change the options.

Configure Traffic Marking through ACL Rules

You can mark traffic for egress packets through iptables or ip6tables rule classifications. To enable these rules, you do one of the following:

To enable traffic marking, use cl-acltool. Add the -p option to specify the location of the policy file. By default, if you do not include the -p option, cl-acltool looks for the policy file in /etc/cumulus/acl/policy.d/.

The iptables-/ip6tables-based marking is supported with the following action extension:

-j SETQOS --set-dscp 10 --set-cos 5

For ebtables, the setqos keyword must be in lowercase, as in:

[ebtables]
-A FORWARD -o swp5 -j setqos --set-cos 5

You can specify one of the following targets for SETQOS/setqos:

OptionDescription
--set-cos INTSets the datapath resource/queuing class value. Values are defined in IEEE P802.1p.
--set-dscp valueSets the DSCP field in packet header to a value, which can be either a decimal or hex value.
--set-dscp-class classSets the DSCP field in the packet header to the value represented by the DiffServ class value. This class can be EF, BE or any of the CSxx or AFxx classes.

You can specify either --set-dscp or --set-dscp-class, but not both.

Here are two example rules:

[iptables]
-t mangle -A FORWARD --in-interface swp+ -p tcp --dport bgp -j SETQOS --set-dscp 10 --set-cos 5

[ip6tables]
-t mangle -A FORWARD --in-interface swp+ -j SETQOS --set-dscp 10

You can put the rule in either the mangle table or the default filter table; the mangle table and filter table are put into separate TCAM slices in the hardware.

To put the rule in the mangle table, include -t mangle; to put the rule in the filter table, omit -t mangle.

Priority Flow Control

Priority flow control, as defined in the IEEE 802.1Qbb standard, provides a link-level flow control mechanism that can be controlled independently for each Class of Service (CoS) with the intention to ensure no data frames are lost when congestion occurs in a bridged network.

PFC is not supported on switches with the Helix4 ASIC.

PFC is a layer 2 mechanism that prevents congestion by throttling packet transmission. When PFC is enabled for received packets on a set of switch ports, the switch detects congestion in the ingress buffer of the receiving port and signals the upstream switch to stop sending traffic. If the upstream switch has PFC enabled for packet transmission on the designated priorities, it responds to the downstream switch and stops sending those packets for a period of time.

PFC operates between two adjacent neighbor switches; it does not provide end-to-end flow control. However, when an upstream neighbor throttles packet transmission, it could build up packet congestion and propagate PFC frames further upstream: eventually the sending server could receive PFC frames and stop sending traffic for a time.

The PFC mechanism can be enabled for individual switch priorities on specific switch ports for RX and/or TX traffic. The switch port’s ingress buffer occupancy is used to measure congestion. If congestion is present, the switch transmits flow control frames to the upstream switch. Packets with priority values that do not have PFC configured are not counted during congestion detection; neither do they get throttled by the upstream switch when it receives flow control frames.

PFC congestion detection is implemented on the switch using xoff and xon threshold values for the specific ingress buffer which is used by the targeted switch priorities. When a packet enters the buffer and the buffer occupancy is above the xoff threshold, the switch transmits an Ethernet PFC frame to the upstream switch to signal packet transmission should stop. When the buffer occupancy drops below the xon threshold, the switch sends another PFC frame upstream to signal that packet transmission can resume. (PFC frames contain a quanta value to indicate a timeout value for the upstream switch: packet transmission can resume after the timer has expired, or when a PFC frame with quanta == 0 is received from the downstream switch.)

After the downstream switch has sent a PFC frame upstream, it continues to receive packets until the upstream switch receives and responds to the PFC frame. The downstream ingress buffer must be large enough to store those additional packets after the xoff threshold has been reached.

Priority flow control is fully supported on both Broadcom and Mellanox switches.

PFC is disabled by default in Cumulus Linux. To enable priority flow control (PFC), you must configure the following settings in the /etc/cumulus/datapath/traffic.conf file on the switch:

The following configuration example shows PFC configured for ports swp1 through swp4 and swp6:

# to configure priority flow control on a group of ports:
# -- assign cos value(s) to the cos list
# -- add or replace a port group names in the port group list
# -- for each port group in the list
#    -- populate the port set, e.g.
#       swp1-swp4,swp8,swp50s0-swp50s3
#    -- set a PFC buffer size in bytes for each port in the group
#    -- set the xoff byte limit (buffer limit that triggers PFC frame transmit to start)
#    -- set the xon byte delta (buffer limit that triggers PFC frame transmit to stop)
#    -- enable PFC frame transmit and/or PFC frame receive
# priority flow control
pfc.port_group_list = [pfc_port_group]
pfc.pfc_port_group.cos_list = []
pfc.pfc_port_group.port_set = swp1-swp4,swp6
pfc.pfc_port_group.port_buffer_bytes = 25000
pfc.pfc_port_group.xoff_size = 10000
pfc.pfc_port_group.xon_delta = 2000
pfc.pfc_port_group.tx_enable = true
pfc.pfc_port_group.rx_enable = true

Port Groups

A port group refers to one or more sequences of contiguous ports. You can define multiple port groups by adding:

You can specify the set of ports in a port group in comma-separate sequences of contiguous ports; you can see which ports are contiguous in the /var/lib/cumulus/porttab file. The syntax supports:

...
swp2
swp3
swp4
swp5
swp6s0
swp6s1
swp6s2
swp6s3
swp7
...

On a Broadcom switch, restart switchd with the sudo systemctl restart switchd.service command to allow the PFC configuration changes to take effect. On a Mellanox switch with the Spectrum ASIC, restarting switchd is not necessary.

The PAUSE frame is a flow control mechanism that halts the transmission of the transmitter for a specified period of time. A server or other network node within the data center may be receiving traffic faster than it can handle it, thus the PAUSE frame. In Cumulus Linux, you can configure individual ports to execute link pause by:

Link pause is disabled by default. To enabling link pause, you must configure settings in the /etc/cumulus/datapath traffic.conf file.

What’s the difference between link pause and priority flow control?

  • Priority flow control is applied to an individual priority group for a specific ingress port.
  • Link pause (also known as port pause or global pause) is applied to all the traffic for a specific ingress port.

Here is an example configuration that enables both types of link pause for swp1 through swp4 and swp6:

# to configure pause on a group of ports:
# -- add or replace port group names in the port group list
# -- for each port group in the list
#    -- populate the port set, e.g.
#       swp1-swp4,swp8,swp50s0-swp50s3
#    -- set a pause buffer size in bytes for each port in the group
#    -- set the xoff byte limit (buffer limit that triggers pause frames transmit to start)
#    -- set the xon byte delta (buffer limit that triggers pause frames transmit to stop)

# link pause
link_pause.port_group_list = [pause_port_group]
link_pause.pause_port_group.port_set = swp1-swp4,swp6
link_pause.pause_port_group.port_buffer_bytes = 25000
link_pause.pause_port_group.xoff_size = 10000
link_pause.pause_port_group.xon_delta = 2000
link_pause.pause_port_group.rx_enable = true
link_pause.pause_port_group.tx_enable = true

On a Broadcom switch, restart switchd with the sudo systemctl restart switchd.service command to allow the PFC configuration changes to take effect. On a Mellanox switch with the Spectrum ASIC, restarting switchd is not necessary.

Cut-through Mode and Store and Forward Switching

Cut-through mode is disabled in Cumulus Linux by default on switches with Broadcom ASICs. With cut-though mode enabled and link pause is asserted, Cumulus Linux generates a TOVR and TUFL ERROR; certain error counters increment on a given physical port.

cumulus@switch:~$ sudo ethtool -S swp49 | grep Error
HwIfInDot3LengthErrors: 0
HwIfInErrors: 0
HwIfInDot3FrameErrors: 0
SoftInErrors: 0
SoftInFrameErrors: 0
HwIfOutErrors: 35495749
SoftOutErrors: 0

cumulus@switch:~$ sudo ethtool -S swp50 | grep Error
HwIfInDot3LengthErrors: 3038098
HwIfInErrors: 297595762
HwIfInDot3FrameErrors: 293710518

To work around this issue, disable link pause or disable cut-through mode in the /etc/cumulus/datapath/traffic.conf file.

To disable link pause, comment out the link_pause* section in the /etc/cumulus/datapath/traffic.conf file:

cumulus@switch:~$ sudo nano /etc/cumulus/datapath/traffic.conf 
#link_pause.port_group_list = [port_group_0]
#link_pause.port_group_0.port_set = swp45-swp54
#link_pause.port_group_0.rx_enable = true
#link_pause.port_group_0.tx_enable = true

To enable store and forward switching, set cut_through_enable to false in the /etc/cumulus/datapath/traffic.conf file:

cumulus@switch:~$ sudo nano /etc/cumulus/datapath/traffic.conf
cut_through_enable = false

On switches using Broadcom Tomahawk, Trident II, Trident II+, and Trident3 ASICs, Cumulus Linux supports store and forward switching but does not support cut-through mode.

On switches with the Mellanox Spectrum ASIC, Cumulus Linux supports cut-through mode but does not support store and forward switching.

Congestion Notification

Explicit Congestion Notification (ECN) is defined by RFC 3168. ECN enables the Cumulus Linux switch to mark a packet to signal impending congestion instead of dropping the packet, which is how TCP typically behaves when ECN is not enabled.

ECN is a layer 3 end-to-end congestion notification mechanism only. Packets can be marked as ECN-capable transport (ECT) by the sending server. If congestion is observed by any switch while the packet is getting forwarded, the ECT-enabled packet can be marked by the switch to indicate the congestion. The end receiver can respond to the ECN-marked packets by signaling the sending server to slow down transmission. The sending server marks a packet ECT by setting the least 2 significant bits in an IP header DiffServ (ToS) field to 01 or 10. A packet that has the least 2 significant bits set to 00 indicates a non-ECT-enabled packet.

The ECN mechanism on a switch only marks packets to notify the end receiver. It does not take any other action or change packet handling in any way, nor does it respond to packets that have already been marked ECN by an upstream switch.

On Trident II switches only, if ECN is enabled on a specific queue, the ASIC also enables RED on the same queue. If the packet is ECT marked (the ECN bits are 01 or 10), the ECN mechanism executes as described above. However, if it is entering an ECN-enabled queue but is not ECT marked (the ECN bits are 00), then the RED mechanism uses the same threshold and probability values to decide whether to drop the packet. Packets entering a non-ECN-enabled queue do not get marked or dropped due to ECN or RED in any case.

ECN is implemented on the switch using minimum and maximum threshold values for the egress queue length. When a packet enters the queue and the average queue length is between the minimum and maximum threshold values, a configurable probability value will determine whether the packet will be marked. If the average queue length is above the maximum threshold value, the packet is always marked.

The downstream switches with ECN enabled perform the same actions as the traffic is received. If the ECN bits are set, they remain set. The only way to overwrite ECN bits is to set the ECN bits to 11.

ECN is supported on Broadcom Tomahawk, Tomahawk2, Trident II, Trident II+ and Trident3, and Mellanox Spectrum ASICs.

Click to learn how to configure ECN

ECN is disabled by default in Cumulus Linux. You can enable ECN for individual switch priorities on specific switch ports in the /etc/cumulus/datapath/traffic.conf file:

  • Specify the name of the port group in ecn.port_group_list in brackets; for example, ecn.port_group_list = [ecn_port_group].
  • Assign a CoS value to the port group in ecn.ecn_port_group.cos_list. If the CoS value of a packet matches the value of this setting, then ECN is applied. Note that ecn_port_group is the name of a port group you specified above.
  • Populate the port group with its member ports (ecn.ecn_port_group.port_set), where ecn_port_group is the name of the port group you specified above. Congestion is measured on the egress port queue for the ports listed here, using the average queue length: if congestion is present, a packet entering the queue may be marked to indicate that congestion was observed. Marking a packet involves setting the least 2 significant bits in the IP header DiffServ (ToS) field to 11.
  • The switch priority value(s) are mapped to specific egress queues for the target switch ports.
  • The ecn.ecn_port_group.probability value indicates the probability of a packet being marked if congestion is experienced.

The following configuration example shows ECN configured for ports swp1 through swp4 and swp6:

# Explicit Congestion Notification
# to configure ECN on a group of ports:
# -- add or replace port group names in the port group list
# -- assign cos value(s) to the cos list  *ECN will only be applied to traffic matching this COS*
# -- for each port group in the list
#    -- populate the port set, e.g.
#       swp1-swp4,swp8,swp50s0-swp50s3
  ecn.port_group_list = [ecn_port_group]
  ecn.ecn_port_group.cos_list = [0]
  ecn.ecn_port_group.port_set = swp1-swp4,swp6
  ecn.ecn_port_group.min_threshold_bytes = 40000
  ecn.ecn_port_group.max_threshold_bytes = 200000
  ecn.ecn_port_group.probability = 100

On a Broadcom switch, restart switchd with the sudo systemctl restart switchd.service command to allow the PFC configuration changes to take effect. On a Mellanox switch with the Spectrum ASIC, restarting switchd is not necessary.

Check Interface Buffer Status

iptables-extensions man page

Hardware-enabled DDOS Protection

It is crucial to protect the control plane on the switch to ensure that the proper control plane applications have access to the CPU. Failure to do so increases vulnerabilities to a Denial of Service (DOS attack. Cumulus Linux provides control plane protection by default. In addition, you can configure DDOS protection to protect data plane, control plane, and management plane traffic on the switch. You can configure Cumulus Linux to drop packets that match one or more of the following criteria while incurring no performance impact:

DDOS protection is not supported on Broadcom Hurricane2 and Mellanox Spectrum ASICs.

Configure DDOS Protection

  1. Open the /etc/cumulus/datapath/traffic.conf file in a text editor.

  2. Enable DOS prevention checks by setting the dos_enable value to true:

# To turn on/off Denial of Service (DOS) prevention checks
dos_enable = true
  1. Open the /usr/lib/python2.7/dist-packages/cumulus/__chip_config/bcm/datapath.conf file in a text editor. Set any of the DOS checks to true. For example:
cumulus@switch:~$ sudo nano /usr/lib/python2.7/dist-packages/cumulus/__chip_config/bcm/datapath.conf
# Enabling/disabling Denial of service (DOS) prevetion checks
# To change the default configuration:
# enable/disable the individual DOS checks.
dos.sip_eq_dip = true
dos.smac_eq_dmac = true
dos.tcp_hdr_partial = true
dos.tcp_syn_frag = true
dos.tcp_ports_eq = true
dos.tcp_flags_syn_fin = true
dos.tcp_flags_fup_seq0 = true
dos.tcp_offset1 = true
dos.tcp_ctrl0_seq0 = true
dos.udp_ports_eq = true
dos.icmp_frag = true
dos.icmpv4_length = true
dos.icmpv6_length = true
dos.ipv6_min_frag = true

Configuring any of the following settings affects the BFD echo function. For example, if you enable dos.udp_ports_eq, all the BFD packets are dropped because the BFD protocol uses the same source and destination UDP ports.

dos.sip_eq_dip
dos.smac_eq_dmac
dos.tcp_ctrl0_seq0
dos.tcp_flags_fup_seq0
dos.tcp_flags_syn_fin
dos.tcp_ports_eq
dos.tcp_syn_frag
dos.udp_ports_eq

  1. Restart switchd:
cumulus@switch:~$ sudo systemctl restart switchd.service

DHCP Relays

DHCP is a client/server protocol that automatically provides IP hosts with IP addresses and other related configuration information. A DHCP relay (agent) is a host that forwards DHCP packets between clients and servers. DHCP relays forward requests and replies between clients and servers that are not on the same physical subnet.

This topic describes how to configure DHCP relays for IPv4 and IPv6. Configurations on the server hosts, DHCP relays, and DHCP server are provided using the following topology:

The dhcpd and dhcrelay services are disabled by default. After you finish configuring the DHCP relays and servers, you need to start those services. If you intend to run these services within a VRF, including the management VRF, follow these steps.

Configure IPv4 DHCP Relays

To configure IPv4 DHCP relays, run the following commands.

NCLU Commands

You configure a DHCP relay on a per-VLAN basis, specifying the SVI, not the parent bridge. In the example below, you specify vlan1 as the SVI for VLAN 1 but you do not specify the bridge named bridge in this case.

Specify the IP address of each DHCP server and the interfaces that are used as the uplinks. In the example commands below, the DHCP server IP address is 172.16.1.102, VLAN 1 (the SVI is vlan1) and the uplinks are swp51 and swp52. As per RFC 3046, you can specify as many server IP addresses that can fit in 255 octets. You can specify each address only once.

cumulus@switch:~$ net add dhcp relay interface swp51
cumulus@switch:~$ net add dhcp relay interface swp52
cumulus@switch:~$ net add dhcp relay interface vlan1
cumulus@switch:~$ net add dhcp relay server 172.16.1.102
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

These commands create the following configuration in the /etc/default/isc-dhcp-relay file:

SERVERS="172.16.1.102"
INTF_CMD="-i vlan1 -i swp51 -i swp52"
OPTIONS=""

Enable, then restart the dhcrelay service so the configuration persists between reboots:

cumulus@switch:~$ sudo systemctl enable dhcrelay.service
cumulus@switch:~$ sudo systemctl restart dhcrelay.service
Linux Commands
  1. Edit the /etc/default/isc-dhcp-relay file to add the IP address of the DHCP server and both interfaces participating in DHCP relay (facing the server and facing the client). In the example below, the DHCP server IP address is 172.16.1.102, VLAN 1 (the SVI is vlan1) and the uplinks are swp51 and swp52. If the client-facing interface is a bridge port, specify the switch virtual interface (SVI) name if using a VLAN-aware bridge (for example, bridge.100), or the bridge name if using traditional bridging (for example, br100).
cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
SERVERS="172.16.1.102"
INTF_CMD="-i vlan1 -i swp51 -i swp52"
OPTIONS=""
  1. Enable then restart the dhcrelay service so that the configuration persists between reboots:
cumulus@switch:~$ sudo systemctl enable dhcrelay.service
cumulus@switch:~$ sudo systemctl restart dhcrelay.service

To see the DHCP relay status, use the systemctl status dhcrelay.service command:

cumulus@switch:~$ sudo systemctl status dhcrelay.service
● dhcrelay.service - DHCPv4 Relay Agent Daemon
    Loaded: loaded (/lib/systemd/system/dhcrelay.service; enabled)
    Active: active (running) since Fri 2016-12-02 17:09:10 UTC; 2min 16s ago
      Docs: man:dhcrelay(8)
Main PID: 1997 (dhcrelay)
    CGroup: /system.slice/dhcrelay.service
            └─1997 /usr/sbin/dhcrelay --nl -d -q -i vlan1 -i swp51 -i swp52 172.16.1.102

DHCP Agent Information Option (Option 82)

Cumulus Linux supports DHCP Agent Information Option 82, which allows a DHCP relay to insert circuit or relay specific information into a request that is being forwarded to a DHCP server. Two sub-options are provided:

To enable the DHCP Agent Information Option, you configure the -a option. By default, when you enable this option, the Circuit ID is the printable name of the interface on which the client request is received, typically an SVI. The Remote ID is the System MAC of the device on which DHCP relay is running.

NCLU commands are not currently available for this feature. Use the following Linux commands.

cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-a"
cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-a --use-pif-circuit-id" 
cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-a -r CUSTOMVALUE"

Make sure to restart the dhcrelay service to apply the new configuration :

cumulus@switch:~$ sudo systemctl restart dhcrelay.service

Control the Gateway IP Address with RFC 3527

When DHCP relay is required in an environment that relies on an anycast gateway (such as EVPN), a unique IP address is necessary on each device for return traffic. By default, in a BGP unnumbered environment with DHCP relay, the source IP address is set to the loopback IP address and the gateway IP address (giaddr) is set as the SVI IP address. However with anycast traffic, the SVI IP address is not unique to each rack; it is typically shared amongst all racks. Most EVPN ToR deployments only possess a single unique IP address, which is the loopback IP address.

RFC 3527 enables the DHCP server to react to these environments by introducing a new parameter to the DHCP header called the link selection sub-option, which is built by the DHCP relay agent. The link selection sub-option takes on the normal role of the giaddr in relaying to the DHCP server which subnet is correlated to the DHCP request. When using this sub-option, the giaddr continues to be present but only relays the return IP address that is to be used by the DHCP server; the giaddr becomes the unique loopback IP address.

When enabling RFC 3527 support, you can specify an interface, such as the loopback interface or a switch port interface to be used as the giaddr. The relay picks the first IP address on that interface. If the interface has multiple IP addresses, you can specify a specific IP address for the interface.

RFC 3527 is supported for IPv4 DHCP relays only.

The following illustration demonstrates how you can control the giaddr with RFC 3527.

To enable RFC 3527 support and control the giaddr, run the following commands.

NCLU Commands
  1. Run the net add dhcp relay giaddr-interface command with the interface/IP address you want to use. The following example uses the first IP address on the loopback interface as the giaddr:
cumulus@switch:~$ net add dhcp relay giaddr-interface lo

The above command creates the following configuration in the /etc/default/isc-dhcp-relay file:

# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U lo"

The first IP address on the loopback interface is typically the 127.0.0.1 address; Cumulus Networks recommends that you use more specific syntax, as shown in the next example.

The following example uses IP address 10.0.0.1 on the loopback interface as the giaddr:

cumulus@switch:~$ net add dhcp relay giaddr-interface lo 10.0.0.1

The above command creates the following configuration in the /etc/default/isc-dhcp-relay file:

# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U 10.0.0.1%lo"

The following example uses the first IP address on swp2 as the giaddr:

cumulus@switch:~$ net add dhcp relay giaddr-interface swp2

The above command creates the following configuration in the /etc/default/isc-dhcp-relay file:

# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U swp2"

The following example uses IP address 10.0.0.3 on swp2 as the giaddr:

cumulus@switch:~$ net add dhcp relay giaddr-interface swp2 10.0.0.3

The above command creates the following configuration in the /etc/default/isc-dhcp-relay file:

# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U 10.0.0.3%swp2"
  1. Restart the dhcrelay service to apply the configuration change:
cumulus@switch:~$ sudo systemctl restart dhcrelay.service
Linux Commands
  1. Edit the /etc/default/isc-dhcp-relay file and provide the -U option with the interface or IP address you want to use as the giaddr. The following example uses the first IP address on the loopback interface as the giaddr:
cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U lo"

The first IP address on the loopback interface is typically the 127.0.0.1 address; Cumulus Networks recommends that you use more specific syntax, as shown in the next example.

The following example uses IP address 10.0.0.1 on the loopback interface as the giaddr:

cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U 10.0.0.1%lo"

The following example uses the first IP address on swp2 as the giaddr:

cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U swp2"

The following example uses IP address 10.0.0.3 on swp2 as the giaddr:

cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
...
# Additional options that are passed to the DHCP relay daemon?
OPTIONS="-U 10.0.0.3%swp2"
  1. Restart the dhcrelay service to apply the configuration change :
cumulus@switch:~$ sudo systemctl restart dhcrelay.service

Gateway IP Address as Source IP for Relayed DHCP Packets (Advanced)

You can configure the dhcrelay service to forward IPv4 (only) DHCP packets to a DHCP server and ensure that the source IP address of the relayed packet is the same as the gateway IP address.

This option impacts all relayed IPv4 packets globally.

To use the gateway IP address as the source IP address:

NCLU Commands

Run these commands:

cumulus@leaf:~$ net add dhcp relay use-giaddr-as-src
cumulus@leaf:~$ net pending
cumulus@leaf:~$ net commit
Linux Commands
  1. Edit the /etc/default/isc-dhcp-relay file to add --giaddr-src to the OPTIONS line. An example is shown below.
cumulus@switch:~$ sudo nano /etc/default/isc-dhcp-relay
SERVERS="172.16.1.102"
INTF_CMD="-i vlan1 -i swp51 -i swp52 -U swp2"
OPTIONS="--giaddr-src"
  1. Restart the dhcrelay service to apply the configuration change :
cumulus@switch:~$ sudo systemctl restart dhcrelay.service

Configure IPv6 DHCP Relays

NCLU commands are not currently available to configure IPv6 relays.

  1. Edit the /etc/default/isc-dhcp-relay6 file to add the upstream and downstream interfaces. In the example below, the SVI is vlan1, and the interfaces are swp51 and swp52.
cumulus@switch:$ sudo nano /etc/default/isc-dhcp-relay6 
SERVERS=" -u 2001:db8:100::2%swp51 -u 2001:db8:100::2%swp52"
INTF_CMD="-l vlan1"
  1. Enable, then restart the dhcrelay6 service so that the configuration persists between reboots:
cumulus@switch:~$ sudo systemctl enable dhcrelay6.service
cumulus@switch:~$ sudo systemctl restart dhcrelay6.service

To see the status of the IPv6 DHCP relay, use the systemctl status dhcrelay6.service command:

cumulus@switch:~$ sudo systemctl status dhcrelay6.service
● dhcrelay6.service - DHCPv6 Relay Agent Daemon
    Loaded: loaded (/lib/systemd/system/dhcrelay6.service; disabled)
    Active: active (running) since Fri 2016-12-02 21:00:26 UTC; 1s ago
      Docs: man:dhcrelay(8)
  Main PID: 6152 (dhcrelay)
    CGroup: /system.slice/dhcrelay6.service
            └─6152 /usr/sbin/dhcrelay -6 --nl -d -q -l vlan1 -u 2001:db8:100::2 swp51 -u 2001:db8:100::2 swp52

Configure Multiple DHCP Relays

Cumulus Linux supports multiple DHCP relay daemons on a switch to enable relaying of packets from different bridges to different upstream interfaces.

To configure multiple DHCP relay daemons on a switch:

  1. As the sudo user, open the /etc/vrf/systemd.conf file in a text editor and remove dhcrelay.
  2. Run the following command to reload the systemd files:
cumulus@switch:~$ sudo systemctl daemon-reload
  1. Create a configuration file in the /etc/default directory for each DHCP relay daemon. Use the naming scheme isc-dhcp-relay-<dhcp-name> for IPv4 or isc-dhcp-relay6-<dhcp-name> for IPv6. An example configuration file for IPv4 is shown below:
# Defaults for isc-dhcp-relay initscript
# sourced by /etc/init.d/isc-dhcp-relay
# installed at /etc/default/isc-dhcp-relay by the maintainer scripts

#
# This is a POSIX shell fragment
#

# What servers should the DHCP relay forward requests to?
SERVERS="102.0.0.2"
# On what interfaces should the DHCP relay (dhrelay) serve DHCP requests?
# Always include the interface towards the DHCP server.
# This variable requires a -i for each interface configured above.
# This will be used in the actual dhcrelay command
# For example, "-i eth0 -i eth1"
INTF_CMD="-i swp2s2 -i swp2s3"

# Additional options that are passed to the DHCP relay daemon?
OPTIONS=""

An example configuration file for IPv6 is shown below:

# Defaults for isc-dhcp-relay6 initscript
# sourced by /etc/init.d/isc-dhcp-relay6
# installed at /etc/default/isc-dhcp-relay6 by the maintainer scripts

#
# This is a POSIX shell fragment
#

# Specify upstream and downstream interfaces
# For example, "-u eth0 -l swp1"
INTF_CMD=""

# Additional options that are passed to the DHCP relay daemon?
OPTIONS=""
  1. Run the following command to start a dhcrelay instance, where <``dhcp-name> is the instance name or number.
cumulus@switch:~$ sudo systemctl start dhcrelay@<dhcp-name>

Configure a DHCP Relay with VRR

The configuration procedure for DHCP relay with VRR is the same as documented above.

The DHCP relay must run on the SVI and not on the -v0 interface.

Troubleshooting

If you are experiencing issues with DHCP relay, you can check if there is a problem with systemd:

 cumulus@switch:~$  /usr/sbin/dhcrelay -4 -i <interface-facing-host> <ip-address-dhcp-server> -i <interface-facing-dhcp-server>
 cumulus@switch:~$  /usr/sbin/dhcrelay -6 -l <interface-facing-host> -u <ip-address-hcp-server>%<interface-facing-dhcp-server>

For example:

cumulus@switch:~$ /usr/sbin/dhcrelay -4 -i vlan1 172.16.1.102 -i swp51
cumulus@switch:~$ /usr/sbin/dhcrelay -6 -l vlan1 -u 2001:db8:100::2%swp51

The above commands manually activate the DHCP relay process and they do not persist when you reboot the switch.

To see how DHCP relay is working on your switch, run the journalctl command:

cumulus@switch:~$ sudo journalctl -l -n 20 | grep dhcrelay
Dec 05 20:58:55 leaf01 dhcrelay[6152]: sending upstream swp52
Dec 05 20:58:55 leaf01 dhcrelay[6152]: sending upstream swp51
Dec 05 20:58:55 leaf01 dhcrelay[6152]: Relaying Reply to fe80::4638:39ff:fe00:3 port 546 down.
Dec 05 20:58:55 leaf01 dhcrelay[6152]: Relaying Reply to fe80::4638:39ff:fe00:3 port 546 down.
Dec 05 21:03:55 leaf01 dhcrelay[6152]: Relaying Renew from fe80::4638:39ff:fe00:3 port 546 going up.
Dec 05 21:03:55 leaf01 dhcrelay[6152]: sending upstream swp52
Dec 05 21:03:55 leaf01 dhcrelay[6152]: sending upstream swp51
Dec 05 21:03:55 leaf01 dhcrelay[6152]: Relaying Reply to fe80::4638:39ff:fe00:3 port 546 down.
Dec 05 21:03:55 leaf01 dhcrelay[6152]: Relaying Reply to fe80::4638:39ff:fe00:3 port 546 down.

To specify a time period with the journalctl command, use the --since flag:

cumulus@switch:~$ sudo journalctl -l --since "2 minutes ago" | grep dhcrelay
Dec 05 21:08:55 leaf01 dhcrelay[6152]: Relaying Renew from fe80::4638:39ff:fe00:3 port 546 going up.
Dec 05 21:08:55 leaf01 dhcrelay[6152]: sending upstream swp52
Dec 05 21:08:55 leaf01 dhcrelay[6152]: sending upstream swp51

Configuration Errors

If you configure DHCP relays by editing the /etc/default/isc-dhcp-relay file manually instead of running NCLU commands, you might introduce configuration errors that can cause the switch to crash.

For example, if you see an error similar to the following, there might be a space between the DHCP server address and the interface used as the uplink.

Core was generated by /usr/sbin/dhcrelay --nl -d -i vx-40 -i vlan100 10.0.0.4 -U 10.0.1.2  %vlan120.
Program terminated with signal SIGSEGV, Segmentation fault.

To resolve the issue, manually edit the /etc/default/isc-dhcp-relay file to remove the space, then run the systemctl restart dhcrelay.service command to restart the dhcrelay service and apply the configuration change.

Caveats and Errata

Interface Names Cannot Be Longer than 14 Characters

The dhcrelay command does not bind to an interface if the interface’s name is longer than 14 characters. To work around this issue, change the interface name to be 14 or fewer characters if dhcrelay is required to bind to it.

This is a known limitation in dhcrelay.

DHCP Servers

A DHCP Server automatically provides and assigns IP addresses and other network parameters to client devices. It relies on the Dynamic Host Configuration Protocol to respond to broadcast requests from clients.

This topic describes how to configure a DHCP server for IPv4 and IPv6. Configurations on the hosts, DHCP relay and DHCP server are provided using the following topology. The DHCP server is a switch running Cumulus Linux; however, the DHCP server can also be located on a dedicated server in your environment.

The dhcpd and dhcrelay services are disabled by default. After you finish configuring the DHCP relays and servers, you need to start those services. If you intend to run these services within a VRF, including the management VRF, follow these steps.

For information about DHCP relays, refer to DHCP Relays.

Configure the DHCP Server on Cumulus Linux Switches

To configure the DHCP server on a Cumulus Linux switch for IPv4 and IPv6, you need to edit the /etc/dhcp/dhcp.conf and /etc/dhcp/dhcpd6.conf configuration files. Sample configurations are provided as a starting point.

You must include two pools in the DHCP configuration files:

Configure the IPv4 DHCP Server

In a text editor, edit the /etc/dhcp/dhcpd.conf file. Use following configuration as an example:

cumulus@switch:~$ cat /etc/dhcp/dhcpd.conf
ddns-update-style none;

default-lease-time 600;
max-lease-time 7200;

subnet 10.0.100.0 netmask 255.255.255.0 {
}
subnet 10.0.1.0 netmask 255.255.255.0 {
      range 10.0.1.50 10.0.1.60;
}

Edit the /etc/default/isc-dhcp-server configuration file so that the DHCP server launches when the system boots. Here is an example configuration:

cumulus@switch:~$ cat /etc/default/isc-dhcp-server
DHCPD_CONF="-cf /etc/dhcp/dhcpd.conf"

INTERFACES="swp1"

Enable and start the dhcpd service:

cumulus@switch:~$ sudo systemctl enable dhcpd.service
cumulus@switch:~$ sudo systemctl start dhcpd.service

Configure the IPv6 DHCP Server

In a text editor, edit the /etc/dhcp/dhcpd6.conf file. Use following configuration as an example:

cumulus@switch:~$ cat /etc/dhcp/dhcpd6.conf
ddns-update-style none;

default-lease-time 600;
max-lease-time 7200;

subnet6 2001:db8:100::/64 {
}
subnet6 2001:db8:1::/64 {
    range6 2001:db8:1::100 2001:db8:1::200;
}

Edit the /etc/default/isc-dhcp-server6 file so that the DHCP server launches when the system boots. Here is an example configuration:

cumulus@switch:~$ cat /etc/default/isc-dhcp-server6
DHCPD_CONF="-cf /etc/dhcp/dhcpd6.conf"

INTERFACES="swp1"

Enable and start the dhcpd6 service:

cumulus@switch:~$ sudo systemctl enable dhcpd6.service
cumulus@switch:~$ sudo systemctl start dhcpd6.service

Assign Port-Based IP Addresses

You can assign an IP address and other DHCP options based on physical location or port regardless of MAC address to clients that are attached directly to the Cumulus Linux switch through a switch port. This is helpful when swapping out switches and servers; you can avoid the inconvenience of collecting the MAC address and sending it to the network administrator to modify the DHCP server configuration.

Edit the /etc/dhcp/dhcpd.conf file and add the interface name ifname to assign an IP address through DHCP. The following provides an example:

host myhost {
    ifname = "swp1" ;
    fixed_address = 10.10.10.10 ;
}

Troubleshooting

The DHCP server determines if a DHCP request is a relay or a non-relay DHCP request. You can run the following command to see the DHCP request:

cumulus@server02:~$ sudo tail /var/log/syslog | grep dhcpd
2016-12-05T19:03:35.379633+00:00 server02 dhcpd: Relay-forward message from 2001:db8:101::1 port 547, link address 2001:db8:101::1, peer address fe80::4638:39ff:fe00:3
2016-12-05T19:03:35.380081+00:00 server02 dhcpd: Advertise NA: address 2001:db8:1::110 to client with duid 00:01:00:01:1f:d8:75:3a:44:38:39:00:00:03 iaid = 956301315 valid for 600 seconds
2016-12-05T19:03:35.380470+00:00 server02 dhcpd: Sending Relay-reply to 2001:db8:101::1 port 547

802.1X Interfaces

The IEEE 802.1X protocol provides a method of authenticating a client (called a supplicant) over wired media. It also provides access for individual MAC addresses on a switch (called the authenticator) after those MAC addresses have been authenticated by an authentication server, typically a RADIUS (Remote Authentication Dial In User Service, defined by RFC 2865) server.

A Cumulus Linux switch acts as an intermediary between the clients connected to the wired ports and the authentication server, which is reachable over the existing network. EAPOL (Extensible Authentication Protocol (EAP) over LAN - EtherType value of 0x888E, defined by RFC 3748) operates on top of the data link layer; the switch uses EAPOL to communicate with supplicants connected to the switch ports.

Cumulus Linux implements 802.1X through the Debian hostapd package, which has been modified to provide the PAE (port access entity).

Supported Features and Limitations

Configure the RADIUS Server

Before you configure any interfaces for 802.1X, configure the RADIUS server.

Do not use a Cumulus Linux switch as the RADIUS server.

To add a popular and freely available RADIUS server called FreeRADIUS on a Debian server, do the following:

root@radius:~# apt-get update
root@radius:~# apt-get install freeradius-utils freeradius-common

When installed and configured, the FreeRADIUS server can serve Cumulus Linux running hostapd as a RADIUS client.

For more information, see the FreeRADIUS documentation.

Configure 802.1X Interfaces

All the 802.1X interfaces share the same RADIUS server settings. Make sure you configure the RADIUS server before you configure the 802.1X interfaces. See Configure the RADIUS Server above.

To configure an 802.1X interface, you need to set the following parameters, then enable 802.1X on the interface:

Configure 802.1X Interfaces for a VLAN-aware Bridge

NCLU Commands

NCLU handles all the 802.1X interface configuration, updating hostapd and other components so you do not have to manually modify configuration files.

  1. Create a simple interface bridge configuration on the switch and add the switch ports that are members of the bridge. You can use glob syntax to add a range of interfaces. The MAB and parking VLAN configurations require interfaces to be bridge access ports. The VLAN-aware bridge must be named bridge and there can be only one VLAN-aware bridge on a switch.
cumulus@switch:~$ net add bridge bridge ports swp1-4
  1. Add the 802.1X RADIUS server IP address and shared secret:
cumulus@switch:~$ net add dot1x radius server-ip 127.0.0.1
cumulus@switch:~$ net add dot1x radius shared-secret mysecret

You can specify a VRF for outgoing RADIUS accounting and authorization packets. The following example specifies a VRF called blue:

cumulus@switch:~$ net add dot1x radius server-ip 127.0.0.1 vrf blue
cumulus@switch:~$ net add dot1x radius shared-secret mysecret
  1. Enable 802.1X on the interfaces, then review and commit the new configuration:
cumulus@switch:~$ net add interface swp1-4 dot1x
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To assign a tagged VLAN for voice devices and assign different VLANs to the devices based on authorization, run these commands:

cumulus@switch:~$ net add interface swp1-4 dot1x voice-enable
cumulus@switch:~$ net add interface swp1-4 dot1x voice-enable vlan 200
cumulus@switch:~$ net pending 
cumulus@switch:~$ net commit
Linux Commands
  1. Edit the /etc/network/interfaces file to create a simple interface bridge configuration on the switch and add the switch ports that are members of the bridge. The MAB and parking VLAN configurations require interfaces to be bridge access ports. The VLAN-aware bridge must be named bridge and there can be only one VLAN-aware bridge on a switch. The following example shows that swp1 thru swp4 are members of the bridge.
cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3 swp4
    bridge-vlan-aware yes
...
  1. Edit the /etc/hostapd.conf file to configure 802.1X settings. The example below sets:
    • The IP address of the 802.1X RADIUS server to 127.0.0.1 (auth_server_addr=127.0.0.1). You can specify a VRF for outgoing RADIUS accounting and authorization packets (for example, to specify a VRF called blue: auth_server_addr=127.0.0.1%blue).
    • The shared secret to mysecret (auth_server_shared_secret=mysecret).
    • 802.1X on swp1 thru swp4 (interfaces=swp1,swp2,swp3,swp4).
cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
interfaces=swp1,swp2,swp3,swp4
mab_interfaces=
parking_vlan_interfaces=
parking_vlan_id=
mab_activation_delay=30
eap_reauth_period=0
eap_send_identity=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=1812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=1813
acct_server_shared_secret=mysecret
...
  1. Enable then restart the hostapd service so that the configuration persists between reboots:
cumulus@switch:~$ sudo systemctl enable hostapd
cumulus@switch:~$ sudo systemctl restart hostapd

Configure 802.1X Interfaces for a Traditional Mode Bridge

NCLU Commands

NCLU and hostapd might change traditional mode configurations on the bridge-ports line in the /etc/network/interface file by adding or deleting special 802.1X traditional mode bridge-ports configuration stanzas in /etc/network/interfaces.d/. The source configuration command in /etc/network/interfaces must include these special configuration filenames. It must include at least source /etc/network/interfaces.d/*.intf so that these files are sourced during an ifreload.

  1. Create uplink ports. The following example uses bonds:
cumulus@switch:~$ net add bond bond1 bond slaves swp5-6
cumulus@switch:~$ net add bond bond2 bond slaves swp7-8
  1. Create a traditional mode bridge configuration on the switch and add the switch ports that are members of the bridge. A traditional bridge cannot be named **** bridge as that name is reserved for the single VLAN-aware bridge on the switch. You can use glob syntax to add a range of interfaces.
cumulus@switch:~$ net add bridge bridge1 ports swp1-4
  1. Create bridge associations with the parking VLAN ID and the dynamic VLAN IDs. In this example, 600 is used for the parking VLAN ID and 700 is used for the dynamic VLAN ID:
cumulus@switch:~$ net add bridge br-vlan600 ports bond1.600
cumulus@switch:~$ net add bridge br-vlan700 ports bond2.700
  1. Add the 802.1X RADIUS server IP address and shared secret:
cumulus@switch:~$ net add dot1x radius server-ip 127.0.0.1
cumulus@switch:~$ net add dot1x radius shared-secret mysecret

You can specify a VRF for outgoing RADIUS accounting and authorization packets.The following example specifies a VRF called blue:

cumulus@switch:~$ net add dot1x radius server-ip 127.0.0.1 vrf blue
cumulus@switch:~$ net add dot1x radius shared-secret mysecret
  1. Enable 802.1X on the interfaces, then review and commit the new configuration:
cumulus@switch:~$ net add interface swp1-2 dot1x
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands
  1. Edit the /etc/network/interfaces file to create uplink ports and create a traditional mode bridge configuration on the switch.

    a. Create uplink ports. The following example uses bonds:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
       bond-slaves swp5 swp6
auto bond2
iface bond2
       bond-slaves swp7 swp8
...

b. Create a traditional mode bridge configuration on the switch and add the switch ports that are members of the bridge. You must also create bridge associations with the parking VLAN ID and the dynamic VLAN IDs. In this example, 600 is used for the parking VLAN ID and 700 is used for the dynamic VLAN ID.

A traditional bridge cannot be named **** bridge as that name is reserved for the single VLAN-aware bridge on the switch. You can use glob syntax to add a range of interfaces.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge1
iface bridge1
      bridge-ports swp1-swp4
      bridge-vlan-aware no

auto br-vlan600
iface br-vlan600
      bridge-ports bond1.600
      bridge-vlan-aware no

auto br-vlan700
iface br-vlan700
      bridge-ports bond1.700
      bridge-vlan-aware no
  1. Edit the /etc/hostapd.conf file to configure 802.1X settings. The example below sets:

    • The IP address of the 802.1X RADIUS server to 127.0.0.1 (auth_server_addr=127.0.0.1). You can specify a VRF for outgoing RADIUS accounting and authorization packets (for example, to specify a VRF called blue: auth_server_addr=127.0.0.1%blue).
    • The shared secret to mysecret (auth_server_shared_secret=mysecret).
    • 802.1X on swp1, swp2, swp3, and swp4 (interfaces=swp1,swp2,swp3,swp4).
cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
interfaces=swp1,swp2,swp3,swp4
mab_interfaces=
parking_vlan_interfaces=
parking_vlan_id=
mab_activation_delay=30
eap_reauth_period=0
eap_send_identity=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=1812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=1813
acct_server_shared_secret=testing123
...
  1. Enable then restart the hostapd service so that the configuration persists between reboots:
cumulus@switch:~$ sudo systemctl enable hostapd
cumulus@switch:~$ sudo systemctl restart hostapd

Configure the Linux Supplicants

A sample FreeRADIUS server configuration needs to contain the entriesfor users host1 and host2 on swp1 and swp2 for them to be placed in a VLAN.

host1 Cleartext-Password := "host1password"
host2 Cleartext-Password := "host2password"

After being configured, each supplicant needs the proper credentials:

user@host1:~# cat /etc/wpa_supplicant.conf

ctrl_interface=/var/run/wpa_supplicant
ctrl_interface_group=0
eapol_version=2
ap_scan=0
network={
        key_mgmt=IEEE8021X
        eap=TTLS MD5
        identity="host1"
        anonymous_identity="host1"
        password="host1password"
        phase1="auth=MD5"
        eapol_flags=0
}
user@host2:~# cat /etc/wpa_supplicant.conf

ctrl_interface=/var/run/wpa_supplicant
ctrl_interface_group=0
eapol_version=2
ap_scan=0
network={
        key_mgmt=IEEE8021X
        eap=TTLS MD5
        identity="host2"
        anonymous_identity="host2"
        password="host2password"
        phase1="auth=MD5"
        eapol_flags=0
  }

To test that a supplicant (client) can communicate with the Cumulus Linux Authenticator switch, run the following command from the supplicant:

root@host1:/home/cumulus# wpa_supplicant -c /etc/wpa_supplicant.conf -D wired -i swp1
Successfully initialized wpa_supplicant
swp1: Associated with 01:80:c2:00:00:03
swp1: CTRL-EVENT-EAP-STARTED EAP authentication started
swp1: CTRL-EVENT-EAP-PROPOSED-METHOD vendor=0 method=4
swp1: CTRL-EVENT-EAP-METHOD EAP vendor 0 method 4 (MD5) selected
swp1: CTRL-EVENT-EAP-SUCCESS EAP authentication completed successfully
swp1: CTRL-EVENT-CONNECTED - Connection to 01:80:c2:00:00:03 compl

Or from another supplicant:

root@host2:/home/cumulus# wpa_supplicant -c /etc/wpa_supplicant.conf -D wired -i swp1
Successfully initialized wpa_supplicant
swp1: Associated with 01:80:c2:00:00:03
swp1: CTRL-EVENT-EAP-STARTED EAP authentication started
swp1: CTRL-EVENT-EAP-PROPOSED-METHOD vendor=0 method=4
swp1: CTRL-EVENT-EAP-METHOD EAP vendor 0 method 4 (MD5) selected
swp1: CTRL-EVENT-EAP-SUCCESS EAP authentication completed successfully
swp1: CTRL-EVENT-CONNECTED - Connection to 01:80:c2:00:00:03 comp

Configure Accounting and Authentication Ports

You can configure the accounting and authentication ports in Cumulus Linux. The default values are 1813 for the accounting port and 1812 for the authentication port. You can also change the reauthentication period for Extensible Authentication Protocol (EAP). The period defaults to 0 (no re-authentication is performed by the switch).

To use different ports:

NCLU Commands

The following example commands change:

  • The authentication port to 2812
  • The accounting port to 2813
  • The reauthentication period for EAP to 86400
cumulus@switch:~$ net add dot1x radius authentication-port 2812
cumulus@switch:~$ net add dot1x radius accounting-port 2813
cumulus@switch:~$ net add dot1x eap-reauth-period 86400
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/hostapd.conf file to change the accounting and authentication ports.The example below sets:

  • The accounting port to 2813 (auth_server_port=2813)
  • The authentication port to 2812
  • The reauthentication period for EAP to 86400 (eap_reauth_period=86400)
cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
eap_reauth_period=86400
eap_send_identity=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=2812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=2813
...

Restart the hostapd service :

cumulus@switch:~$ sudo systemctl restart hostapd

Configure MAC Authentication Bypass

MAC authentication bypass (MAB) enables bridge ports to allow devices to bypass authentication based on their MAC address. This is useful for devices that do not support PAE, such as printers or phones. You can change the MAB activation delay from the default of 30 seconds, but the delay must be between 5 and 30 seconds. After the delay limit is reached, the port enters MAB mode.

MAB must be configured on both the RADIUS server and the RADIUS client (the Cumulus Linux switch).

When using a VLAN-aware bridge, the switch port must be part of bridge named bridge.

To configure MAB:

NCLU Commands

Enable a bridge port for MAB. The following example commands enable bridge port swp1 for MAB:

cumulus@switch:~$ net add interface swp1 dot1x mab
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/hostapd.conf file to enable a bridge port for MAB. The following example enables bridge port swp1 for MAB.

cumulus@switch:~$ sudo nano hostapd.conf
...
mab_interfaces=swp1
parking_vlan_interfaces=
parking_vlan_id=
mab_activation_delay=20
...

Restart the hostapd service:

cumulus@switch:~$ sudo systemctl restart hostapd

Configure a Parking VLAN

If a non-authorized supplicant tries to communicate with the switch, you can route traffic from that device to a different VLAN and associate that VLAN with one of the switch ports to which the supplicant is attached.

For VLAN-aware bridges, the parking VLAN is assigned by manipulating the PVID of the switch port. For traditional mode bridges, Cumulus Linux identifies the bridge associated with the parking VLAN ID and moves the switch port into that bridge. If an appropriate bridge is not found for the move, the port remains in an unauthenticated state where no packets can be received or transmitted.

When using a VLAN-aware bridge, the switch port must be part of bridge named bridge.

NCLU Commands

Run the following commands:

cumulus@switch:~$ net add dot1x parking-vlan-id 777
cumulus@switch:~$ net add interface swp1 dot1x parking-vlan
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

If the authentication for swp1 fails, the port is moved to the parking VLAN:

cumulus@switch:~$ net show dot1x interface swp1 details

Interface  MAC Address        Attribute                     Value
---------  -----------------  ----------------------------  -----------------
swp1       00:02:00:00:00:08  Status Flags                  [PARKED_VLAN]
                              Username                      vlan60
                              Authentication Type           MD5
                              VLAN                          777
                              Session Time (seconds)        24772
                              EAPOL Frames RX               9
                              EAPOL Frames TX               12
                              EAPOL Start Frames RX         1
                              EAPOL Logoff Frames RX        0
                              EAPOL Response ID Frames RX   4
                              EAPOL Response Frames RX      8
                              EAPOL Request ID Frames TX    4
                              EAPOL Request Frames TX       8
                              EAPOL Invalid Frames RX       0
                              EAPOL Length Error Frames Rx  0
                              EAPOL Frame Version           2
                              EAPOL Auth Last Frame Source  00:02:00:00:00:08
                              EAPOL Auth Backend Responses  8
                              RADIUS Auth Session ID        C2FED91A39D8D605

The following output shows a parking VLAN association failure. A VLAN association failure only occurs with traditional mode bridges when there is no traditional bridge available with a parking VLAN ID-tagged subinterface (notice the [UNKNOWN_BR] status in the output):

cumulus@switch:~$ net show dot1x interface swp3 details

Interface  MAC Address        Attribute                     Value
---------  -----------------  ----------------------------  -------------------------
swp1       00:02:00:00:00:08  Status Flags                  [PARKED_VLAN][UNKNOWN_BR]
                              Username                      vlan60
                              Authentication Type           MD5
                              VLAN                          777
                              Session Time (seconds)        24599
                              EAPOL Frames RX               3
                              EAPOL Frames TX               3
                              EAPOL Start Frames RX         1
                              EAPOL Logoff Frames RX        0
                              EAPOL Response ID Frames RX   1
                              EAPOL Response Frames RX      2
                              EAPOL Request ID Frames TX    1
                              EAPOL Request Frames TX       2
                              EAPOL Invalid Frames RX       0
                              EAPOL Length Error Frames Rx  0
                              EAPOL Frame Version           2
                              EAPOL Auth Last Frame Source  00:02:00:00:00:08
                              EAPOL Auth Backend Responses  2
                              RADIUS Auth Session ID        C2FED91A39D8D605
Linux Commands

Edit the /etc/hostapd.conf file to add the parking VLAN ID and port. The following example adds the parking VLAN ID 777 (parking_vlan_id=777) and port swp1 (parking_vlan_interfaces=swp1)

cumulus@switch:~$ sudo nano hostapd.conf
...
parking_vlan_interfaces=swp1
parking_vlan_id=777
...

If the authentication for swp1 fails, the port is moved to the parking VLAN.

Configure Dynamic VLAN Assignments

A common requirement for campus networks is to assign dynamic VLANs to specific users in combination with IEEE 802.1x. After authenticating a supplicant, the user is assigned a VLAN based on the RADIUS configuration.

For VLAN-aware bridges, the dynamic VLAN is assigned by manipulating the PVID of the switch port. For traditional mode bridges, Cumulus Linux identifies the bridge associated with the dynamic VLAN ID and moves the switch port into that bridge. If an appropriate bridge is not found for the move, the port remains in an unauthenticated state where no packets can be received or transmitted.

To enable dynamic VLAN assignment globally, where VLAN attributes sent from the RADIUS server are applied to the bridge:

NCLU Commands

Run the following commands:

cumulus@switch:~$ net add dot1x dynamic-vlan
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

You can specify the require option in the command so that VLAN attributes are required. If VLAN attributes do not exist in the access response packet returned from the RADIUS server, the user is not authorized and has no connectivity. If the RADIUS server returns VLAN attributes but the user has an incorrect password, the user is placed in the parking VLAN (if you have configured parking VLAN).

cumulus@switch:~$ net add dot1x dynamic-vlan require
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The following example shows a typical RADIUS configuration (shown for FreeRADIUS, not typically configured or run on the Cumulus Linux device) for a user with dynamic VLAN assignment:

# # VLAN 100 Client Configuration for Freeradius RADIUS Server.
# # This is not part of the CL configuration.
vlan100client Cleartext-Password := "client1password"
      Service-Type = Framed-User,
      Tunnel-Type = VLAN,
      Tunnel-Medium-Type = "IEEE-802",
      Tunnel-Private-Group-ID = 100

Verify the configuration (notice the [AUTHORIZED] status in the output):

cumulus@switch:~$ net show dot1x interface swp1 details

Interface  MAC Address        Attribute                     Value
---------  -----------------  ----------------------------  --------------------------
swp1       00:02:00:00:00:08  Status Flags                  [DYNAMIC_VLAN][AUTHORIZED]
                              Username                      host1
                              Authentication Type           MD5
                              VLAN                          888
                              Session Time (seconds)        799
                              EAPOL Frames RX               3
                              EAPOL Frames TX               3
                              EAPOL Start Frames RX         1
                              EAPOL Logoff Frames RX        0
                              EAPOL Response ID Frames RX   1
                              EAPOL Response Frames RX      2
                              EAPOL Request ID Frames TX    1
                              EAPOL Request Frames TX       2
                              EAPOL Invalid Frames RX       0
                              EAPOL Length Error Frames Rx  0
                              EAPOL Frame Version           2
                              EAPOL Auth Last Frame Source  00:02:00:00:00:08
                              EAPOL Auth Backend Responses  2
                              RADIUS Auth Session ID        939B1A53B624FC56
cumulus@switch:~$ net show dot1x interface summary

Interface  MAC Address        Username      State         Authentication Type  MAB  VLAN
---------  -----------------  ------------  ------------  -------------------  ---  ----
swp1       00:02:00:00:00:08  000200000008  AUTHORIZED    unknown              NO   888

The following output shows a dynamic VLAN association failure. VLAN association failure only occurs with traditional mode bridges when there is no traditional bridge available with a parking VLAN ID-tagged subinterface in it (notice the [UNKNOWN_BR] status in the output):

cumulus@switch:~$ net show dot1x interface swp1 details

Interface  MAC Address        Attribute                     Value
---------  -----------------  ----------------------------  --------------------------------------
swp1       00:02:00:00:00:08  Status Flags                  [DYNAMIC_VLAN][AUTHORIZED][UNKNOWN_BR]
                              Username                      host2
                              Authentication Type           MD5
                              VLAN                          888
                              Session Time (seconds)        11
                              EAPOL Frames RX               3
                              EAPOL Frames TX               3
                              EAPOL Start Frames RX         1
                              EAPOL Logoff Frames RX        0
                              EAPOL Response ID Frames RX   1
                              EAPOL Response Frames RX      2
                              EAPOL Request ID Frames TX    1
                              EAPOL Request Frames TX       2
                              EAPOL Invalid Frames RX       0
                              EAPOL Length Error Frames Rx  0
                              EAPOL Frame Version           2
                              EAPOL Auth Last Frame Source  00:02:00:00:00:08
                              EAPOL Auth Backend Responses  2
                              RADIUS Auth Session ID        BDF731EF2B765B78
Linux Commands

Edit the /etc/hostapd.conf file to add the following options:

  • dynamic_vlan=1 (Specify dynamic_vlan=2 if you want VLAN attributes to be required. If VLAN attributes do not exist in the access response packet returned from the RADIUS server, the user is not authorized and has no connectivity. If the RADIUS server returns VLAN attributes but the user has an incorrect password, the user is placed in the parking VLAN, if you have configured parking VLAN).
  • radius_das_port=
  • radius_das_time_window=300
  • radius_das_require_event_timestamp=1
  • radius_das_require_message_authenticator=1

Remove the eap_send_identity=0 option. For example:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
eap_server=0
ieee8021x=1
driver=wired
dynamic_vlan=1
interfaces=swp1,swp2,swp3,swp4
mab_interfaces=
parking_vlan_interfaces=swp1
parking_vlan_id=777
mab_activation_delay=30
eap_reauth_period=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=1812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=1813
acct_server_shared_secret=
radius_das_port=
radius_das_time_window=300
radius_das_require_event_timestamp=1
radius_das_require_message_authenticator=1

Restart the hostapd service :

cumulus@switch:~$ sudo systemctl restart hostapd

The following example shows a typical RADIUS configuration (shown for FreeRADIUS, not typically configured or run on the Cumulus Linux device) for a user with dynamic VLAN assignment:

# # VLAN 100 Client Configuration for Freeradius RADIUS Server.
# # This is not part of the CL configuration.
vlan100client Cleartext-Password := "client1password"
      Service-Type = Framed-User,
      Tunnel-Type = VLAN,
      Tunnel-Medium-Type = "IEEE-802",
      Tunnel-Private-Group-ID = 100

To disable dynamic VLAN assignment, where VLAN attributes sent from the RADIUS server are ignored and users are authenticated based on existing credentials:

NCLU Commands

Run the net del dot1x dynamic-vlan command:

cumulus@switch:~$ net del dot1x dynamic-vlan
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/hostapd.conf file to remove the following options:

  • dynamic_vlan=1
  • radius_das_port=
  • radius_das_time_window=300
  • radius_das_require_event_timestamp=1
  • radius_das_require_message_authenticator=1

Add the eap_send_identity=0 option. The following example shows the options in the /etc/hostapd.conf file

cumulus@switch:~$ sudo nano /etc/hostapd.conf
eap_server=0
ieee8021x=1
driver=wired
interfaces=
mab_interfaces=
parking_vlan_interfaces=swp1
parking_vlan_id=777
mab_activation_delay=30
eap_reauth_period=0
eap_send_identity=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=1812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=1813
acct_server_shared_secret=

Restart the hostapd service.

cumulus@switch:~$ sudo systemctl restart hostapd

Enabling or disabling dynamic VLAN assignment restarts hostapd, which forces existing, authorized users to re-authenticate.

Configure MAC Addresses per Port

You can specify the maximum number of authenticated MAC addresses allowed on a port.

NCLU Commands

Run the net add dot1x max-number-stations <value> command. You can specify any number between 0 and 255. The default value is 4.

cumulus@switch:~$ net add dot1x max-number-stations 10
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/hostapd.conf file to add the max_number_stations= option. For example:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
eap_server=0
ieee8021x=1
driver=wired
dynamic_vlan=1
max_number_stations=10
interfaces=swp1,swp2,swp3,swp4
mab_interfaces=
parking_vlan_interfaces=swp1
parking_vlan_id=777
mab_activation_delay=30
eap_reauth_period=0
ctrl_interface=/var/run/hostapd
nas_identifier=localhost
auth_server_addr=127.0.0.1
auth_server_port=1812
auth_server_shared_secret=mysecret
acct_server_addr=
acct_server_port=1813
acct_server_shared_secret=
radius_das_port=
radius_das_time_window=300
radius_das_require_event_timestamp=1
radius_das_require_message_authenticator=1

Restart the hostapd service :

cumulus@switch:~$ sudo systemctl restart hostapd

Configure EAP Requests from the Switch

Cumulus Linux provides the send-eap-request-id option, which you can use to trigger EAP packets to be sent from the host side of a connection. For example, this option is required in a configuration where a PC connected to a phone attempts to send EAP packets to the switch via the phone but the PC does not receive a response from the switch (the phone might not be ready to forward packets to the switch after a reboot). Because the switch does not receive EAP packets, it attempts to authorize the PC with MAB instead of waiting for the packets. In this case, the PC might be placed into a parking VLAN to isolate it. To remove the PC from the parking VLAN, the switch needs to send an EAP request to the PC to trigger EAP.

To configure the switch send an EAP request, run these commands:

cumulus@switch:~$ net add dot1x send-eap-request-id
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Only run this command if MAB is configured on an interface.

The PC might attempt 802.1X authorization through the bridged connection in the back of the phone before the phone completes MAB authorization. In this case, 802.1X authorization fails.

The net del dot1x send-eap-request-id command disables this feature.

RADIUS Change of Authorization and Disconnect Requests

Extensions to the RADIUS protocol (RFC 5176) enable the Cumulus Linux switch to act as a Dynamic Authorization Server (DAS) by listening for Change of Authorization (CoA) requests from the RADIUS server (Dynamic Authorization Client (DAC)) and taking action when needed, such as bouncing a port or terminating a user session. The IEEE 802.1x server (hostapd) running on Cumulus Linux has been adapted to handle these additional, unsolicited RADIUS requests.

Configure DAS

To configure DAS, provide the UDP port (3799 is the default port), the IP address, and the secret key for the DAS client.

NCLU Commands

The following example commands set the UDP port to the default port, the IP address of the DAS client to 10.0.2.228, and the secret key to myclientsecret:

cumulus@switch:~$ net add dot1x radius das-port default
cumulus@switch:~$ net add dot1x radius das-client-ip 10.0.2.228 das-client-secret mysecret123
cumulus@switch:~$ net commit

You can specify a VRF so that incoming RADIUS disconnect and CoA commands are received and acknowledged on the correct interface when VRF is configured. The following example specifies VRF blue:

cumulus@switch:~$ net add dot1x radius das-port default
cumulus@switch:~$ net add dot1x radius das-client-ip 10.0.2.228 vrf blue das-client-secret mysecret123
cumulus@switch:~$ net commit

You can configure up to four DAS clients to be authorized to send CoA commands. For example:

cumulus@switch:~$ net add dot1x radius das-port default
cumulus@switch:~$ net add dot1x radius das-client-ip 10.20.250.53 das-client-secret mysecret1
cumulus@switch:~$ net add dot1x radius das-client-ip 10.0.1.7 das-client-secret mysecret2
cumulus@switch:~$ net add dot1x radius das-client-ip 10.20.250.99 das-client-secret mysecret3
cumulus@switch:~$ net add dot1x radius das-client-ip 10.10.0.0.2 das-client-secret mysecret4
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To see DAS configuration information, run the net show configuration dot1x command. For example:

cumulus@switch:~$ net show configuration dot1x

dot1x
  mab-activation-delay 5
  eap-reauth-period 0
  parking-vlan-id 100
  dynamic-vlan

  radius
    accounting-port 1813
    das-client-ip 10.0.2.228 das-client-secret mysecret123
    authentication-port 1812
    das-port 3799
Linux Commands

Edit the /etc/hostapd.conf file to add the following options to configure the UDP port, the IP address and secret key for the DAS client:

  • radius_das_port
  • radius_das_client

The following example sets the UDP port to the default port, the IP address of the DAS client to 10.0.2.228, and the secret key to mysecret123:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
radius_das_port=3799
radius_das_client=10.0.2.228 mysecret123

You can specify a VRF so that incoming RADIUS disconnect and CoA commands are received and acknowledged on the correct interface when VRF is configured. The following example specifies VRF blue:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
radius_das_port=3799
radius_das_client=10.0.2.228%blue mysecret123

You can configure up to four DAS clients to be authorized to send CoA commands. For example:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
radius_das_port=3799
radius_das_client=10.20.250.53 mysecret1 
radius_das_client=10.0.1.7 mysecret2
radius_das_client=10.20.250.99 mysecret3
radius_das_client=10.10.0.0.2 mysecret4

Restart the hostapd service:

cumulus@switch:~$ sudo systemctl restart hostapd

You can disable DAS in Cumulus Linux at any time by running the following commands:

NCLU Commands
cumulus@switch:~$ net del dot1x radius das-port
cumulus@switch:~$ net del dot1x radius das-client-ip
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/hostapd.conf file to remove the following options:

  • radius_das_port
  • radius_das_client

Restart the hostapd service:

cumulus@switch:~$ sudo systemctl restart hostapd

Terminate a User Session

From the DAC, users can create a disconnect message using the radclient utility (included in the Debian freeradius-utils package) on the RADIUS server or other authorized client. A disconnect message is sent as an unsolicited RADIUS Disconnect-Request packet to the switch to terminate a user session and discard all associated session context. The Disconnect-Request packet is used when the RADIUS server wants to disconnect the user after the session has been accepted by the RADIUS Access-Accept packet.

This is an example of a disconnect message created using the radclient utility:

$ echo "Acct-Session-Id=D91FE8E51802097" > disconnect-packet.txt
$ ## OPTIONAL ## echo "User-Name=somebody" >> disconnect-packet.txt
$ echo "Message-Authenticator=1" >> disconnect-packet.txt
$ echo "Event-Timestamp=1532974019" >> disconnect-packet.txt
# now send the packet with the radclient utility (from freeradius-utils deb package)
$ cat disconnect-packet.txt | radclient -x 10.0.0.1:3799 disconnect myclientsecret

To prevent unauthorized servers from disconnecting users, the Disconnect-Request packet must include certain identification attributes (described below). For a session to be disconnected, all parameters must match their expected values at the switch. If the parameters do not match, the switch discards the Disconnect-Request packet and sends a Disconnect-NAK (negative acknowledgment message).

RADIUS DAS: Acct-Session-Id match
RADIUS DAS: No matches remaining after User-Name check
hostapd_das_find_global_sta: checking ifname=swp2
RADIUS DAS: No matches remaining after Acct-Session-Id check
RADIUS DAS: No matching session found
DAS: Session not found for request from 10.10.0.1:58385
DAS: Reply to 10.10.0.1:58385

The following is an example of the Disconnect-Request packet received by the switch:

RADIUS Protocol
Code: Disconnect-Request (40)
Packet identifier: 0x4f (79)
Length: 53
Authenticator: c0e1fa75fdf594a1cfaf35151a43c6a7
Attribute Value Pairs
AVP: t=Acct-Session-Id(44) l=17 val=D91FE8E51802097
AVP: t=User-Name(1) l=10 val=somebody
AVP: t=Message-Authenticator(80) l=18 val=38cb3b6896623b4b7d32f116fa976cdc
AVP: t=Event-Timestamp(55) l=6 val=1532974019
AVP: t=NAS-IP-Address(4) l=6 val=10.0.0.1

Bounce a Port

You can create a CoA bounce-host-port message from the RADIUS server using the radclient utility (included in the Debian freeradius-utils package). The bounce port can cause a link flap on an authentication port, which triggers DHCP renegotiation from one or more hosts connected to the port.

The following is an example of a Cisco AVPair CoA bounce-host-port message sent from the radclient utility:

$ echo "Acct-Session-Id=D91FE8E51802097" > bounce-packet.txt
$ ## OPTIONAL ## echo "User-Name=somebody" >> bounce-packet.txt
$ echo "Message-Authenticator=1" >> bounce-packet.txt
$ echo "Event-Timestamp=1532974019" >> bounce-packet.txt
$ echo "cisco-avpair='subscriber:command=bounce-host-port' " >> bounce-packet.txt
$ cat bounce-packet.txt | radclient -x 10.0.0.1:3799 coa myclientsecret

The message received by the switch is:

RADIUS Protocol
Code: CoA-Request (43)
Packet identifier: 0x3a (58)
Length: 96
Authenticator: 6480d710802329269d5cae6a59bcfb59
Attribute Value Pairs
AVP: t=Acct-Session-Id(44) l=17 val=D91FE8E51802097
Type: 44
Length: 17
Acct-Session-Id: D91FE8E51802097
AVP: t=User-Name(1) l=10 val=somebody
Type: 1
Length: 10
User-Name: somebody
AVP: t=NAS-IP-Address(4) l=6 val=10.0.0.1
Type: 4
Length: 6
NAS-IP-Address: 10.0.0.1
AVP: t=Vendor-Specific(26) l=43 vnd=ciscoSystems(9)
Type: 26
Length: 43
Vendor ID: ciscoSystems (9)
VSA: t=Cisco-AVPair(1) l=37 val=subscriber:command=bounce-host-port
Type: 1
Length: 37
Cisco-AVPair: subscriber:command=bounce-host-port

Configure the NAS IP Address

You can send the NAS IPv4 or IPv6 address in access request and accounting packets. You can only configure one NAS IP address on the switch, which is used for all interface authorizations.

To configure the NAS IP address, run the following commands:

NCLU Commands

The following command example sets the NAS IP address to 10.0.0.1:

cumulus@switch:~$ net add dot1x radius nas-ip-address 10.0.0.1
Linux Commands

Edit the /etc/hostapd.conf file and configure the own_ip_addr setting with the NAS IP address:

cumulus@switch:~$ sudo nano /etc/hostapd.conf
...
interfaces=swp1,swp2,swp3,swp4
mab_interfaces=
parking_vlan_interfaces=
parking_vlan_id=
mab_activation_delay=30
eap_reauth_period=0
eap_send_identity=0
ctrl_interface=/var/run/hostapd
own_ip_addr=10.0.0.1

Enable, then restart the hostapd service so that the configuration persists between reboots:

cumulus@switch:~$ sudo systemctl enable hostapd
cumulus@switch:~$ sudo systemctl restart hostapd

To delete the NAS IP address, either run the NCLU net del dot1x radius nas-ip-address command or edit the /etc/hostapd.conf file.

Troubleshooting

To check connectivity between two supplicants, ping one host from the other:

root@host1:/home/cumulus# ping 198.150.0.2
PING 11.0.0.2 (11.0.0.2) 56(84) bytes of data.
64 bytes from 11.0.0.2: icmp_seq=1 ttl=64 time=0.604 ms
64 bytes from 11.0.0.2: icmp_seq=2 ttl=64 time=0.552 ms
^C
--- 11.0.0.2 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1000ms
rtt min/avg/max/mdev = 0.552/0.578/0

You can run net show dot1x with the following options for more data:

To check to see which MAC addresses have been authorized by RADIUS:

cumulus@switch:~$ net show dot1x macs
Interface     Attribute   Value
-----------   -------------  -----------------
swp1          MAC Addresses  00:02:00:00:00:01
swp2          No Data
swp3          No Data
swp4          No Data

To check the port detail counters:

cumulus@switch:~$ net show dot1x port-details

Interface  Attribute                                  Value
-----------  ----------------------------------------     ---------
swp1        Mac Addresses                              00:02:00:00:00:01
            authMultiSessionId                         96703ADC82D77DF2
            connected_time                             182
            dot1xAuthEapolFramesRx                     3
            dot1xAuthEapolFramesTx                     3
            dot1xAuthEapolLogoffFramesRx               0
            dot1xAuthEapolReqFramesTx                  2
            dot1xAuthEapolReqIdFramesTx                1
            dot1xAuthEapolRespFramesRx                 2
            dot1xAuthEapolRespIdFramesRx               1
            dot1xAuthEapolStartFramesRx                1
            dot1xAuthInvalidEapolFramesRx              0
            dot1xAuthLastEapolFrameSource              00:02:00:00:00:01
            dot1xAuthLastEapolFrameVersion             2
            dot1xAuthPaeState                          5
            dot1xAuthQuietPeriod                       60
            dot1xAuthReAuthEnabled                     FALSE
            dot1xAuthReAuthPeriod                      0
            dot1xAuthServerTimeout                     30
            dot1xAuthSessionAuthenticMethod            1
            dot1xAuthSessionId                         1B50FE8939FD9F5E
            dot1xAuthSessionTerminateCause             999
            dot1xAuthSessionTime                       182
            dot1xAuthSessionUserName                   testing
            dot1xPaePortProtocolVersion                2
            last_eap_type_as                           4 (MD5)
            last_eap_type_sta                          4 (MD5)

To check RADIUS counters:

cumulus@switch:~$ net show dot1x radius-details swp1

Interface  Attribute                                   Value
-----------  ----------------------------------------    ---------
swp1       radiusAccClientRequests                      1
           radiusAccClientResponses                     1
           radiusAccClientServerPortNumber              1813
           radiusAccServerAddress                       127.0.0.1
           radiusAuthClientAccessAccepts                1
           radiusAuthClientAccessChallenges             1
           radiusAuthClientAccessRejects                0
           radiusAuthClientAccessRequests               0
           radiusAuthClientServerPortNumber             1812
           radiusAuthServerAddress                      127.0.0.1
           radiusAuthServerIndex                        1

...

You can also check logging with the journalctl command:

cumulus@switch:~$ sudo journalctl -f -u hostapd
Apr 19 22:17:11 switch hostapd[12462]: swp1: interface state UNINITIALIZED->ENABLED
Apr 19 22:17:11 switch hostapd[12462]: swp1: AP-ENABLED
Apr 19 22:17:11 switch hostapd[12462]: Reading rule file /etc/cumulus/acl/policy.d/00control_ps ...
Apr 19 22:17:11 switch hostapd[12462]: Processing rules in file /etc/cumulus/acl/policy.d/00...
Apr 19 22:17:12 switch hostapd[12462]: Reading rule file /etc/cumulus/acl/policy.d/100_dot1x...
Apr 19 22:17:12 switch hostapd[12462]: Processing rules in file /etc/cumulus/acl/policy.d/ ..
Apr 19 22:17:12 switch hostapd[12462]: Reading rule file /etc/cumulus/acl/policy.d/99control
Apr 19 22:17:12 switch hostapd[12462]: Processing rules in file /etc/cumulus/acl/policy.d/99
Apr 19 22:17:12 switch hostapd[12462]: Installing acl policy
Apr 19 22:17:12 switch hostapd[12462]: done. 

You can perform more advanced troubleshooting with the following commands.

To increase the debug level in hostapd, copy over the hostapd service file, then add -d, -dd or -ddd to the ExecStart line in the hostapd.service file:

cumulus@switch:~$ cp /lib/systemd/system/hostapd.service /etc/systemd/system/hostapd.service
cumulus@switch:~$ sudo nano /etc/systemd/system/hostapd.service
...
ExecStart=/usr/sbin/hostapd -ddd -c /etc/hostapd.conf
...

To watch debugs with journalctl as supplicants attempt to connect:

cumulus@switch:~$ sudo journalctl -n 1000  -u hostapd      # see the last 1000 lines of hostapd debug logging
cumulus@switch:~$ sudo journalctl -f -u hostapd            # continuous tail of the hostapd daemon debug logging

To check ACL rules in /etc/cumulus/acl/policy.d/100_dot1x_swpX.rules before and after a supplicant attempts to authenticate:

cumulus@switch:~$ sudo cl-acltool -L eb | grep swpXX
cumulus@switch:~$ sudo cl-netstat | grep swpXX           # look at interface counters

To check tc rules in /var/lib/hostapd/acl/tc_swpX.rules with:

cumulus@switch:~$ sudo tc -s filter show dev swpXX parent 1:
cumulus@switch:~$ sudo tc -s filter show dev swpXX parent ffff:
Campus design feature setup (6-part blog series)

Prescriptive Topology Manager - PTM

In data center topologies, right cabling is a time-consuming endeavor and is error prone. Prescriptive Topology Manager (PTM) is a dynamic cabling verification tool to help detect and eliminate such errors. It takes a Graphviz-DOT specified network cabling plan (something many operators already generate), stored in a topology.dot file, and couples it with runtime information derived from LLDP to verify that the cabling matches the specification. The check is performed on every link transition on each node in the network.

You can customize the topology.dot file to control ptmd at both the global/network level and the node/port level.

PTM runs as a daemon, named ptmd.

For more information, see man ptmd(8).

Supported Features

Configure PTM

ptmd verifies the physical network topology against a DOT-specified network graph file, /etc/ptm.d/topology.dot.

PTM supports undirected graphs.

At startup, ptmd connects to lldpd, the LLDP daemon, over a Unix socket and retrieves the neighbor name and port information. It then compares the retrieved port information with the configuration information that it read from the topology file. If there is a match, it is a PASS, else it is a FAIL.

PTM performs its LLDP neighbor check using the PortID ifname TLV information.

Basic Topology Example

This is a basic example DOT file and its corresponding topology diagram. Use the same topology.dot file on all switches and do not split the file per device; this allows for easy automation by pushing/pulling the same exact file on each device.

graph G {
    "spine1":"swp1" -- "leaf1":"swp1";
    "spine1":"swp2" -- "leaf2":"swp1";
    "spine2":"swp1" -- "leaf1":"swp2";
    "spine2":"swp2" -- "leaf2":"swp2";
    "leaf1":"swp3" -- "leaf2":"swp3";
    "leaf1":"swp4" -- "leaf2":"swp4";
    "leaf1":"swp5s0" -- "server1":"eth1";
    "leaf2":"swp5s0" -- "server2":"eth1";
}

ptmd Scripts

ptmd executes scripts at /etc/ptm.d/if-topo-pass and /etc/ptm.d/if-topo-failfor each interface that goes through a change and runs if-topo-pass when an LLDP or BFD check passes or if-topo-fails when the check fails. The scripts receive an argument string that is the result of the ptmctl command, described in the ptmd commands below.

Modify these default scripts as needed.

Configuration Parameters

You can configure ptmd parameters in the topology file. The parameters are classified as host-only, global, per-port/node and templates.

Host-only Parameters

Host-only parameters apply to the entire host on which PTM is running. You can include the hostnametype host-only parameter, which specifies if PTM uses only the host name (hostname) or the fully-qualified domain name (fqdn) while looking for the self-node in the graph file. For example, in the graph file below PTM ignores the FQDN and only looks for switch04 because that is the host name of the switch on which it is running:

Cumulus Networks recommends you always wrap the hostname in double quotes; for example, "www.example.com" to prevent ptmd from failing.

To avoid errors when starting the ptmd process, make sure that /etc/hosts and /etc/hostname both reflect the hostname you are using in the topology.dot file.

graph G {
          hostnametype="hostname"
          BFD="upMinTx=150,requiredMinRx=250"
          "cumulus":"swp44" -- "switch04.cumulusnetworks.com":"swp20"
          "cumulus":"swp46" -- "switch04.cumulusnetworks.com":"swp22"
}

In this next example, PTM compares using the FQDN and looks for switch05.cumulusnetworks.com, which is the FQDN of the switch ion which it is running:

graph G {
          hostnametype="fqdn"
          "cumulus":"swp44" -- "switch05.cumulusnetworks.com":"swp20"
          "cumulus":"swp46" -- "switch05.cumulusnetworks.com":"swp22"
}

Global Parameters

Global parameters apply to every port listed in the topology file. There are two global parameters: LLDP and BFD. LLDP is enabled by default; if no keyword is present, default values are used for all ports. However, BFD is disabled if no keyword is present, unless there is a per-port override configured. For example:

graph G {
          LLDP=""
          BFD="upMinTx=150,requiredMinRx=250,afi=both"
          "cumulus":"swp44" -- "qct-ly2-04":"swp20"
          "cumulus":"swp46" -- "qct-ly2-04":"swp22"
}

Per-port Parameters

Per-port parameters provide finer-grained control at the port level. These parameters override any global or compiled defaults. For example:

graph G {
          LLDP=""
          BFD="upMinTx=300,requiredMinRx=100"
          "cumulus":"swp44" -- "qct-ly2-04":"swp20" [BFD="upMinTx=150,requiredMinRx=250,afi=both"]
          "cumulus":"swp46" -- "qct-ly2-04":"swp22"
}

Templates

Templates provide flexibility in choosing different parameter combinations and applying them to a given port. A template instructs ptmd to reference a named parameter string instead of a default one. There are two parameter strings ptmd supports:

For example:

graph G {
          LLDP=""
          BFD="upMinTx=300,requiredMinRx=100"
          BFD1="upMinTx=200,requiredMinRx=200"
          BFD2="upMinTx=100,requiredMinRx=300"
          LLDP1="match_type=ifname"
          LLDP2="match_type=portdescr"
          "cumulus":"swp44" -- "qct-ly2-04":"swp20" [BFD="bfdtmpl=BFD1", LLDP="lldptmpl=LLDP1"]
          "cumulus":"swp46" -- "qct-ly2-04":"swp22" [BFD="bfdtmpl=BFD2", LLDP="lldptmpl=LLDP2"]
          "cumulus":"swp46" -- "qct-ly2-04":"swp22"
}

In this template, LLDP1 and LLDP2 are templates for LLDP parameters. BFD1 and BFD2 are templates for BFD parameters.

Supported BFD and LLDP Parameters

ptmd supports the following BFD parameters:

The following is an example of a topology with BFD applied at the port level:

graph G {
          "cumulus-1":"swp44" -- "cumulus-2":"swp20" [BFD="upMinTx=300,requiredMinRx=100,afi=v6"]
          "cumulus-1":"swp46" -- "cumulus-2":"swp22" [BFD="detectMult=4"]
}

ptmd supports the following LLDP parameters:

The following is an example of a topology with LLDP applied at the port level:

graph G {
          "cumulus-1":"swp44" -- "cumulus-2":"swp20" [LLDP="match_hostname=fqdn"]
          "cumulus-1":"swp46" -- "cumulus-2":"swp22" [LLDP="match_type=portdescr"]
}

When you specify match_hostname=fqdn, ptmd will match the entire FQDN, (cumulus-2.domain.com in the example below). If you do not specify anything for match_hostname, ptmd matches based on hostname only, (cumulus-3 below), and ignores the rest of the URL:

graph G { 
          "cumulus-1":"swp44" -- "cumulus-2.domain.com":"swp20" [LLDP="match_hostname=fqdn"] 
          "cumulus-1":"swp46" -- "cumulus-3":"swp22" [LLDP="match_type=portdescr"] 
}

Bidirectional Forwarding Detection (BFD)

BFD provides low overhead and rapid detection of failures in the paths between two network devices. It provides a unified mechanism for link detection over all media and protocol layers. Use BFD to detect failures for IPv4 and IPv6 single or multihop paths between any two network devices, including unidirectional path failure detection. For information about configuring BFD using PTM, see BFD.

The FRRouting routing suite enables additional checks to ensure that routing adjacencies are formed only on links that have connectivity conformant to the specification, as determined by ptmd.

You only need to do this to check link state; you do not need to enable PTM to determine BFD status.

When the global ptm-enable option is enabled, every interface has an implied ptm-enable line in the configuration stanza in the interfaces file.

To enable the global ptm-enable option, run the following FRRouting command:

cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# ptm-enable
switch(config)# end
switch# write memory
switch# exit
cumulus@switch:~$ 

To disable the checks, delete the ptm-enable parameter from the interface:

NCLU Commands
cumulus@switch:~$ net del interface swp51 ptm-enable
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
FRR Commands
cumulus@switch:~$ sudo vtysh
switch# conf t
switch(config)# interface swp51
switch(config-if)# no ptm-enable
switch(config-if)# end
switch# write memory
switch# exit
cumulus@switch:~$

If you need to re-enable PTM for that interface:

NCLU Commands
cumulus@switch:~$ net add interface swp51 ptm-enable
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
FRR Commands
cumulus@switch:~$ sudo vtysh
switch# conf t
switch(config)# interface swp51
switch(config-if)# ptm-enable

switch(config-if)# end
switch# write memory
switch# exit
cumulus@switch:~$

With PTM enabled on an interface, the zebra daemon connects to ptmd over a Unix socket. Any time there is a change of status for an interface, ptmd sends notifications to zebra. Zebra maintains a ptm-status flag per interface and evaluates routing adjacency based on this flag. To check the per-interface ptm-status:

NCLU Commands
cumulus@switch:~$ net show interface swp1

Interface swp1 is up, line protocol is up
  Link ups:       0    last: (never)
  Link downs:     0    last: (never)
  PTM status: disabled
  vrf: Default-IP-Routing-Table
  index 3 metric 0 mtu 1550 
  flags: <UP,BROADCAST,RUNNING,MULTICAST>
  HWaddr: c4:54:44:bd:01:41
FRR Commands
switch# show interface swp1
Interface swp1 is up, line protocol is up
  Link ups:       0    last: (never)
  Link downs:     0    last: (never)
  PTM status: disabled
  vrf: Default-IP-Routing-Table
  index 3 metric 0 mtu 1550 
  flags: <UP,BROADCAST,RUNNING,MULTICAST>
  HWaddr: c4:54:44:bd:01:41
...

ptmd Service Commands

PTM sends client notifications in CSV format.

To start or restart the ptmd service, run the following command. The topology.dot file must be present for the service to start.

cumulus@switch:~$ sudo systemctl start|restart|force-reload ptmd.service

To instruct ptmd to read the topology.dot file again to apply the new configuration to the running state without restarting:

cumulus@switch:~$ sudo systemctl reload ptmd.service

To stop the ptmd service:

cumulus@switch:~$ sudo systemctl stop ptmd.service

To retrieve the current running state of ptmd:

cumulus@switch:~$ sudo systemctl status ptmd.service

ptmctl Commands

ptmctl is a client of ptmd that retrieves the operational state of the ports configured on the switch and information about BFD sessions from ptmd. ptmctl parses the CSV notifications sent by ptmd. See man ptmctl for more information.

ptmctl Examples

The examples below contain the following keywords in the output of the cbl status column:

cbl status KeywordDefinition
passThe interface is defined in the topology file, LLDP information is received on the interface, and the LLDP information for the interface matches the information in the topology file.
failThe interface is defined in the topology file, LLDP information is received on the interface, and the LLDP information for the interface does not match the information in the topology file.
N/AThe interface is defined in the topology file, but no LLDP information is received on the interface. The interface might be down or disconnected, or the neighbor is not sending LLDP packets.
The N/A and fail status might indicate a wiring problem to investigate.
The N/A status is not shown when you use the -l option with ptmctl; only interfaces that are receiving LLDP information are shown.

For basic output, use ptmctl without any options:

cumulus@switch:~$ sudo ptmctl

-------------------------------------------------------------
port  cbl     BFD     BFD                  BFD    BFD
      status  status  peer                 local  type
-------------------------------------------------------------
swp1  pass    pass    11.0.0.2             N/A    singlehop
swp2  pass    N/A     N/A                  N/A    N/A
swp3  pass    N/A     N/A                  N/A    N/A  

For more detailed output, use the -d option:

cumulus@switch:~$ sudo ptmctl -d

--------------------------------------------------------------------------------------
port  cbl    exp     act      sysname  portID  portDescr  match  last    BFD   BFD
      status nbr     nbr                                  on     upd     Type  state  
--------------------------------------------------------------------------------------
swp45 pass   h1:swp1 h1:swp1  h1       swp1    swp1       IfName 5m: 5s  N/A   N/A
swp46 fail   h2:swp1 h2:swp1  h2       swp1    swp1       IfName 5m: 5s  N/A   N/A

#continuation of the output
-------------------------------------------------------------------------------------------------
BFD   BFD       det_mult  tx_timeout  rx_timeout  echo_tx_timeout  echo_rx_timeout  max_hop_cnt
peer  DownDiag
-------------------------------------------------------------------------------------------------
N/A   N/A       N/A       N/A         N/A         N/A              N/A              N/A
N/A   N/A       N/A       N/A         N/A         N/A              N/A              N/A

To return information on active BFD sessions ptmd is tracking, use the -b option:

cumulus@switch:~$ sudo ptmctl -b

----------------------------------------------------------
port  peer        state  local         type       diag

----------------------------------------------------------
swp1  11.0.0.2    Up     N/A           singlehop  N/A  
N/A   12.12.12.1  Up     12.12.12.4    multihop   N/A

To return LLDP information, use the -l option. It returns only the active neighbors currently being tracked by ptmd.

cumulus@switch:~$ sudo ptmctl -l

---------------------------------------------
port  sysname  portID  port   match  last
                       descr  on     upd
---------------------------------------------
swp45 h1       swp1    swp1   IfName 5m:59s
swp46 h2       swp1    swp1   IfName 5m:59s 

To return detailed information on active BFD sessions ptmd is tracking, use the -b and -d option (results are for an IPv6-connected peer):

cumulus@switch:~$ sudo ptmctl -b -d

----------------------------------------------------------------------------------------
port  peer                 state  local  type       diag  det   tx_timeout  rx_timeout  
                                                          mult
----------------------------------------------------------------------------------------
swp1  fe80::202:ff:fe00:1  Up     N/A    singlehop  N/A   3     300         900
swp1  3101:abc:bcad::2     Up     N/A    singlehop  N/A   3     300         900

#continuation of output
---------------------------------------------------------------------
echo        echo        max      rx_ctrl  tx_ctrl  rx_echo  tx_echo
tx_timeout  rx_timeout  hop_cnt
---------------------------------------------------------------------
0           0           N/A      187172   185986   0        0
0           0           N/A      501      533      0        0

ptmctl Error Outputs

If there are errors in the topology file or there is no session, PTM returns appropriate outputs. Typical error strings are:

Topology file error [/etc/ptm.d/topology.dot] [cannot find node cumulus] -
please check /var/log/ptmd.log for more info

Topology file error [/etc/ptm.d/topology.dot] [cannot open file (errno 2)] -
please check /var/log/ptmd.log for more info

No Hostname/MgmtIP found [Check LLDPD daemon status] -
please check /var/log/ptmd.log for more info

No BFD sessions . Check connections

No LLDP ports detected. Check connections

Unsupported command

For example:

cumulus@switch:~$ sudo ptmctl
-------------------------------------------------------------------------
cmd         error
-------------------------------------------------------------------------
get-status  Topology file error [/etc/ptm.d/topology.dot]
            [cannot open file (errno 2)] - please check /var/log/ptmd.log 
            for more info

If you encounter errors with the topology.dot file, you can use dot (included in the Graphviz package) to validate the syntax of the topology file.

Open the topology file with Graphviz to ensure that it is readable and that the file format is correct.

If you edit topology.dot file from a Windows system, be sure to double check the file formatting; there might be extra characters that keep the graph from working correctly.

Caveats and Errata

When PTMD is incorrectly in a failure state and the Zebra interface is enabled, PIF BGP sessions do not establish the route, but the subinterface on top of it does establish routes.

If the subinterface is configured on the physical interface and the physical interface is incorrectly marked as being in a PTM FAIL state, routes on the physical interface are not processed in FRR, but the subinterface is working.

Port Security

Port security is a layer 2 traffic control feature that enables you to manage network access from end-users. Use port security to:

You can specify what action to take when there is a port security violation (drop packets or put the port into ADMIN down state) and add a timeout for the action to take effect.

  • Port security is supported on Broadcom switches only.
  • Layer 2 interfaces in trunk or access mode are currently supported. However, interfaces in a bond are not supported.

Configure MAC Address Options

To limit port access to a specific MAC address, run the following commands.

The example commands configure swp1 to allow access to MAC address 00:02:00:00:00:05:

cumulus@switch:~$ net add interface swp1 port-security allowed-mac 00:02:00:00:00:05

You can specify only one MAC address with the NCLU command. To specify multiple MAC addresses, set the interface.<port>.port_security.allowed_mac parameter in the /etc/cumulus/switchd.d/port_security.conf file. See Configure Port Security Manually below.

To enable sticky MAC on a port, where the first learned MAC address on the port is the only MAC address allowed, run the following commands.

You can add a timeout value so that after the time specified, the MAC address ages out and no longer has access to the port. The default aging timeout value is 1800 seconds. You can specify a value between 0 and 3600 seconds.

The example commands enable sticky MAC on interface swp1, set the timeout value to 2000 seconds, and enable aging.

cumulus@switch:~$ net add interface swp1 port-security sticky-mac
cumulus@switch:~$ net add interface swp1 port-security sticky-mac timeout 2000
cumulus@switch:~$ net add interface swp1 port-security sticky-mac aging
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To limit the number of MAC addresses that are allowed to access a port, run the following commands. You can specify a number between 0 and 512. The default is 32.

The example commands configure swp1 to limit access to 40 MAC addresses:

cumulus@switch:~$ net add interface swp1 port-security mac-limit 40
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Configure Security Violation Actions

You can configure the action you want to take when there is a security violation on a port:

You can also set a timeout value between 0 and 3600 seconds for the action to take effect. The default is 1800 seconds.

The following example commands put swp1 into ADMIN down state when there is a security violation and set the timeout value to 3600 seconds:

cumulus@switch:~$ net add interface swp1 port-security violation shutdown
cumulus@switch:~$ net add interface swp1 port-security violation timeout 3600
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Enable Port Security Settings

After you configure the port security settings to suit your needs, you can enable security on a port with the following commands:

cumulus@switch:~$ net add interface swp1 port-security
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

To disable port security on a port, run the net del interface <interface> port-security command.

Configure Port Security Manually

You can edit the /etc/cumulus/switchd.d/port_security.conf file manually to configure port security instead of running the NCLU commands shown above. This procedure is useful if you use configuration scripts.

Add the configuration settings you want to use to the /etc/cumulus/switchd.d/port_security.conf file, then restart switchd to apply the changes.

Setting
Description
interface.<port>.port_security.enable1 enables security on the port. 0 disables security on the port.
interface.<port>.port_security.mac_limitThe maximum number of MAC addresses allowed to access the port. You can specify a number between 0 and 512. The default is 32.
interface.<port>.port_security.allowed_macThe specific MAC addresses allowed to access the port. You can specify multiple MAC addresses. Separate each MAC address with a space.
interface.<port>.port_security.sticky_mac1 enables sticky MAC, where the first learned MAC address on the port is the only MAC address allowed. 0 disables sticky MAC.
interface.<port>.port_security.sticky_timeoutThe time period after which the first learned MAC address ages out and no longer has access to the port. The default aging timeout value is 1800 seconds. You can specify a value between 0 and 3600 seconds.
interface.swp1.port_security.sticky_aging1 enables sticky MAC aging. 0 disables sticky MAC aging.
interface.<port>.port_security.violation_modeThe violation mode: 0 (shutdown) puts a port into ADMIN down state. 1 (restrict) drops packets.
interface.<port>.port_security.violation_timeoutThe number of seconds after which the violation mode times out. You can specify a value between 0 and 3600 seconds. The default value is 1800 seconds.

An example /etc/cumulus/switchd.d/port_security.conf configuration file is shown here:

cumulus@switch:~$ sudo nano /etc/cumulus/switchd.d/port_security.conf
interface.swp1.port_security.enable = 1
interface.swp1.port_security.mac_limit = 32
interface.swp1.port_security.static_mac = 00:02:00:00:00:05 00:02:00:00:00:06
interface.swp1.port_security.sticky_mac = 1
interface.<port>.port_security.sticky_timeout = 2000
interface.swp1.port_security.sticky_aging = 1
interface.swp1.port_security.violation_mode = 0
interface.swp1.port_security.violation_timeout = 3600
...

Show Port Security Configuration

To show port security settings for all ports:

cumulus@switch:~$ net show port-security
Interface  Port security  MAC limit  Sticky MAC  Sticky MAC aging  Sticky MAC timeout  Violation mode  Timeout
---------  -------------  ---------  ----------  ----------------  ------------------  --------------  -------
swp1       ENABLED        40         ENABLED     ENABLED           2000                Shutdown        3600
swp2       Disabled       NA         NA          NA                NA                  Restrict        1800
swp3       Disabled       NA         NA          NA                NA                  Restrict        1800
swp4       Disabled       NA         NA          NA                NA                  Restrict        1800
swp5       Disabled       NA         NA          NA                NA                  Restrict        1800
swp6       Disabled       NA         NA          NA                NA                  Restrict        1800
...

To show port security settings for a specific port:

cumulus@switch:~$ net show port-security swp1
Interface           swp1
Port security       Enabled
Mac limit           40
Sticky mac          ENABLED
Sticky MAC aging    Enabled
Sticky MAC timeout  1440
Violation mode      Shutdown
Violation timeout   3600
Mac addresses
00:02:00:00:00:05
00:02:00:00:00:06

Layer 2

This section describes layer 2 configuration, such as Ethernet bridging, bonding, spanning tree protocol, multi-chassis link aggregation (MLAG), link layer discovery protocol (LLDP), LACP bypass, virtual router redundancy (VRR) and IGMP and MLD snooping.

Spanning Tree and Rapid Spanning Tree

Spanning tree protocol (STP) is always recommended in layer 2 topologies as it prevents bridge loops and broadcast radiation on a bridged network. STP also provides redundant links for automatic failover when an active link fails. STP is enabled by default for both VLAN-aware and traditional bridges.

Supported Modes

Cumulus Linux supports RSTP, PVST, and PVRST modes:

STP for a VLAN-aware Bridge

VLAN-aware bridges only operate in RSTP mode. STP bridge protocol data units (BPDUs) are transmitted on the native VLAN.

If a bridge running RSTP (802.1w) receives a common STP (802.1D) BPDU, it falls back to 802.1D operation automatically. RSTP interoperates with MST seamlessly, creating a single instance of spanning tree, which transmits BPDUs on the native VLAN. RSTP treats the MST domain as one giant switch.

When connecting a VLAN-aware bridge to a proprietary PVST+ switch using STP, VLAN 1 must be allowed on all 802.1Q trunks that interconnect them, regardless of the configured native VLAN. Only VLAN 1 enables the switches to address the BPDU frames to the IEEE multicast MAC address. The proprietary switch might be configured like this:

switchport trunk allowed vlan 1-100

STP for a Traditional Mode Bridge

Per VLAN Spanning Tree (PVST) creates a spanning tree instance for a bridge. Rapid PVST (PVRST) supports RSTP enhancements for each spanning tree instance. To use PVRST with a traditional bridge, you must create a bridge corresponding to the untagged native VLAN and all the physical switch ports must be part of the same VLAN.

For maximum interoperability, when connected to a switch that has a native VLAN configuration, the native VLAN must be configured to be VLAN 1 only.

Show Bridge and STP Status and Logs

To check STP status for a bridge:

NCLU Commands

Run the net show bridge spanning-tree command:

cumulus@switch:~$ net show bridge spanning-tree
bridge CIST info
  enabled         yes
  bridge id       1.000.44:38:39:FF:40:90
  designated root 1.000.44:38:39:FF:40:90
  regional root   1.000.44:38:39:FF:40:90
  root port       none
  path cost     0          internal path cost   0
  max age       20         bridge max age       20
  forward delay 15         bridge forward delay 15
  tx hold count 6          max hops             20
  hello time    2          ageing time          300
  force protocol version     rstp
  time since topology change 253343s
  topology change count      4
  topology change            no
  topology change port       peerlink
  last topology change port  leaf03-04

bridge:exit01-02 CIST info
  enabled            no                      role                 Disabled
  port id            8.004                   state                discarding
  external port cost 305                     admin external cost  0
  internal port cost 305                     admin internal cost  0
  designated root    1.000.44:38:39:00:00:27 dsgn external cost   0
  dsgn regional root 1.000.44:38:39:00:00:27 dsgn internal cost   0
  designated bridge  1.000.44:38:39:00:00:27 designated port      8.004
  admin edge port    no                      auto edge port       yes
  oper edge port     no                      topology change ack  no
  point-to-point     yes                     admin point-to-point auto
  restricted role    no                      restricted TCN       no
  port hello time    2                       disputed             no
  bpdu guard port    no                      bpdu guard error     no
  network port       no                      BA inconsistent      no
  Num TX BPDU        2                       Num TX TCN           0
  Num RX BPDU        0                       Num RX TCN           0
  Num Transition FWD 0                       Num Transition BLK   2
  bpdufilter port    no
  clag ISL           no                      clag ISL Oper UP     no
  clag role          primary                 clag dual conn mac   00:00:00:00:00:00
  clag remote portID F.FFF                   clag system mac      44:38:39:FF:40:90
bridge:leaf01-02 CIST info
  enabled            yes                     role                 Designated
  port id            8.003                   state                forwarding
  external port cost 10000                   admin external cost  0
  internal port cost 10000                   admin internal cost  0
  designated root    1.000.44:38:39:FF:40:90 dsgn external cost   0
  dsgn regional root 1.000.44:38:39:FF:40:90 dsgn internal cost   0
  designated bridge  1.000.44:38:39:FF:40:90 designated port      8.003
  admin edge port    no                      auto edge port       yes
  oper edge port     no                      topology change ack  no
  point-to-point     yes                     admin point-to-point auto
  restricted role    no                      restricted TCN       no
  port hello time    2                       disputed             no
  bpdu guard port    no                      bpdu guard error     no
  network port       no                      BA inconsistent      no
  Num TX BPDU        253558                  Num TX TCN           2
  Num RX BPDU        253373                  Num RX TCN           4
  Num Transition FWD 126675                  Num Transition BLK   126694
  bpdufilter port    no
  clag ISL           no                      clag ISL Oper UP     no
  clag role          primary                 clag dual conn mac   44:38:39:FF:40:94
  clag remote portID F.FFF                   clag system mac      44:38:39:FF:40:90
bridge:leaf03-04 CIST info
  enabled            yes                     role                 Designated
  port id            8.001                   state                forwarding
  external port cost 10000                   admin external cost  0
  internal port cost 10000                   admin internal cost  0
  designated root    1.000.44:38:39:FF:40:90 dsgn external cost   0
  dsgn regional root 1.000.44:38:39:FF:40:90 dsgn internal cost   0
  designated bridge  1.000.44:38:39:FF:40:90 designated port      8.001
  admin edge port    no                      auto edge port       yes
  oper edge port     no                      topology change ack  no
  point-to-point     yes                     admin point-to-point auto
  restricted role    no                      restricted TCN       no
  port hello time    2                       disputed             no
  bpdu guard port    no                      bpdu guard error     no
  network port       no                      BA inconsistent      no
  Num TX BPDU        130960                  Num TX TCN           6
  Num RX BPDU        4                       Num RX TCN           1
  Num Transition FWD 2                       Num Transition BLK   1
  bpdufilter port    no
  clag ISL           no                      clag ISL Oper UP     no
  clag role          primary                 clag dual conn mac   44:38:39:FF:40:93
  clag remote portID F.FFF                   clag system mac      44:38:39:FF:40:90
bridge:peerlink CIST info
  enabled            yes                     role                 Designated
  port id            F.002                   state                forwarding
  external port cost 10000                   admin external cost  0
  internal port cost 10000                   admin internal cost  0
  designated root    1.000.44:38:39:FF:40:90 dsgn external cost   0
  dsgn regional root 1.000.44:38:39:FF:40:90 dsgn internal cost   0
  designated bridge  1.000.44:38:39:FF:40:90 designated port      F.002
  admin edge port    no                      auto edge port       yes
  oper edge port     no                      topology change ack  no
  point-to-point     yes                     admin point-to-point auto
  restricted role    no                      restricted TCN       no
  port hello time    2                       disputed             no
  bpdu guard port    no                      bpdu guard error     no
  network port       no                      BA inconsistent      no
  Num TX BPDU        126700                  Num TX TCN           2
  Num RX BPDU        6                       Num RX TCN           3
  Num Transition FWD 2                       Num Transition BLK   1
  bpdufilter port    no
  clag ISL           yes                     clag ISL Oper UP     yes
  clag role          primary                 clag dual conn mac   00:00:00:00:00:00
  clag remote portID F.FFF                   clag system mac      44:38:39:FF:40:90
Linux Commands

The mstpctl utility provided by the mstpd service configures STP. The mstpd daemon is an open source project used by Cumulus Linux to implement IEEE802.1D 2004 and IEEE802.1Q 2011.

The mstpd daemon starts by default when the switch boots. The mstpd logs and errors are located in /var/log/syslog.

mstpd is the preferred utility for interacting with STP on Cumulus Linux. brctl also provides certain methods for configuring STP; however, they are not as complete as the tools offered in mstpd and output from brctl can be misleading in some cases.

To show the bridge state, run the brctl show command:

cumulus@switch:~$ sudo brctl show
  bridge name     bridge id               STP enabled     interfaces
  bridge          8000.001401010100       yes             swp1
                                                          swp4
                                                          swp5

To show the mstpd bridge port state, run the mstpctl showport bridgecommand:

cumulus@switch:~$ sudo mstpctl showport bridge
  E swp1 8.001 forw F.000.00:14:01:01:01:00 F.000.00:14:01:01:01:00 8.001 Desg
    swp4 8.002 forw F.000.00:14:01:01:01:00 F.000.00:14:01:01:01:00 8.002 Desg
  E swp5 8.003 forw F.000.00:14:01:01:01:00 F.000.00:14:01:01:01:00 8.003 Desg

Customize Spanning Tree Protocol

There are a number of ways to customize STP in Cumulus Linux. Exercise extreme caution with the settings below to prevent malfunctions in STP loop avoidance.

Spanning Tree Priority

If you have a multiple spanning tree instance (MSTI 0, also known as a common spanning tree, or CST), you can set the tree priority for a bridge. The bridge with the lowest priority is elected the root bridge. The priority must be a number between 0 and 61440, and must be a multiple of 4096. The default is 32768.

To set the tree priority, run the following commands:

NCLU Commands

The following example command sets the tree priority to 8192:

cumulus@switch:~$ net add bridge stp treeprio 8192
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Configure the tree priority (mstpctl-treeprio) under the bridge stanza in the /etc/network/interfaces file. The following example command sets the tree priority to 8192:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    # bridge-ports includes all ports related to VxLAN and CLAG.
    # does not include the Peerlink.4094 subinterface
    bridge-ports bond01 bond02 peerlink vni13 vni24 vxlan4001
    bridge-pvid 1
    bridge-vids 13 24
    bridge-vlan-aware yes
    mstpctl-treeprio 8192
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

Cumulus Linux supports MSTI 0 only. It does not support MSTI 1 through 15.

PortAdminEdge (PortFast Mode)

PortAdminEdge is equivalent to the PortFast feature offered by other vendors. It enables or disables the initial edge state of a port in a bridge.

All ports configured with PortAdminEdge bypass the listening and learning states to move immediately to forwarding.

PortAdminEdge mode might cause loops if it is not used with the BPDU guard feature.

It is common for edge ports to be configured as access ports for a simple end host; however, this is not mandatory. In the data center, edge ports typically connect to servers, which might pass both tagged and untagged traffic.

To configure PortAdminEdge mode:

NCLU Commands

The following example commands configure PortAdminEdge and BPDU guard for swp5.

cumulus@switch:~$ net add interface swp5 stp bpduguard
cumulus@switch:~$ net add interface swp5 stp portadminedge
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Configure PortAdminEdge and BPDU guard under the switch port interface stanza in the /etc/network/interfaces file. The following example configures PortAdminEdge and BPD guard on swp5.

cumulus@switch:~$ sudo nano /etc/netowrk/interfaces
...
auto swp5
iface swp5
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To configure PortAdminEdge and BPDU guard at runtime, run the following commands:

cumulus@switch:~$ sudo mstpctl setportadminedge br2 swp1 yes
cumulus@switch:~$ sudo mstpctl setbpduguard br2 swp1 yes

PortAutoEdge

PortAutoEdge is an enhancement to the standard PortAdminEdge (PortFast) mode, which allows for the automatic detection of edge ports. PortAutoEdge enables and disables the auto transition to and from the edge state of a port in a bridge.

Edge ports and access ports are not the same. Edge ports transition directly to the forwarding state and skip the listening and learning stages. Upstream topology change notifications are not generated when an edge port link changes state. Access ports only forward untagged traffic; however, there is no such restriction on edge ports, which can forward both tagged and untagged traffic.

When a BPDU is received on a port configured with PortAutoEdge, the port ceases to be in the edge port state and transitions into a normal STP port. When BPDUs are no longer received on the interface, the port becomes an edge port, and transitions through the discarding and learning states before resuming forwarding.

PortAutoEdge is enabled by default in Cumulus Linux.

To disable PortAutoEdge for an interface:

NCLU Commands

The following example commands disable PortAutoEdge on swp1:

cumulus@switch:~$ net add interface swp1 stp portautoedge no
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the switch port interface stanza in the /etc/network/interfaces file to add the mstpctl-portautoedge no line. The following example disables PortAutoEdge on swp1:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto swp1
iface swp1
    alias to Server01
    # Port to Server02
    mstpctl-portautoedge no
...

Run ifreload -a to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

To re-enable PortAutoEdge for an interface:

NCLU Commands

The following example commands re-enable PortAutoEdge on swp1:

cumulus@switch:~$ net del interface swp1 stp portautoedge no
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the switch port interface stanza in the /etc/network/interfaces file to remove mstpctl-portautoedge no.

Run ifreload -a to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

BPDU Guard

You can configure BPDU guard (Bridge Protocol Data Unit) to protect the spanning tree topology from unauthorized switches affecting the forwarding path. For example, when someone adds a new switch to an access port off a leaf switch and this new switch is configured with a low priority, it might become the new root switch and affect the forwarding path for the entire layer 2 topology.

To configure BPDU guard:

NCLU Commands

The following example commands set BPDU guard for swp5:

cumulus@switch:~$ net add interface swp5 stp bpduguard
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the switch port interface stanza in the /etc/network/interfaces file and add the mstpctl-bpduguard yes line. The following example sets BPDU guard for interface swp5:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto swp5
iface swp5
    mstpctl-bpduguard yes
...

Run ifreload -a to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

If a BPDU is received on the port, STP brings down the port and logs an error in /var/log/syslog. The following is a sample error:

mstpd: error, MSTP_IN_rx_bpdu: bridge:bond0 Recvd BPDU on BPDU Guard Port - Port Down

To determine whether BPDU guard is configured, or if a BPDU has been received:

NCLU Commands
cumulus@switch:~$ net show bridge spanning-tree | grep bpdu
  bpdu guard port    yes                bpdu guard error     yes
Linux Commands
cumulus@switch:~$ mstpctl showportdetail bridge bond0
bridge:bond0 CIST info
  enabled            no                      role                 Disabled
  port id            8.001                   state                discarding
  external port cost 305                     admin external cost  0
  internal port cost 305                     admin internal cost  0
  designated root    8.000.6C:64:1A:00:4F:9C dsgn external cost   0
  dsgn regional root 8.000.6C:64:1A:00:4F:9C dsgn internal cost   0
  designated bridge  8.000.6C:64:1A:00:4F:9C designated port      8.001
  admin edge port    no                      auto edge port       yes
  oper edge port     no                      topology change ack  no
  point-to-point     yes                     admin point-to-point auto
  restricted role    no                      restricted TCN       no
  port hello time    10                      disputed             no
  bpdu guard port    yes                      bpdu guard error     yes
  network port       no                      BA inconsistent      no
  Num TX BPDU        3                       Num TX TCN           2
  Num RX BPDU        488                     Num RX TCN           2
  Num Transition FWD 1                       Num Transition BLK   2
  bpdufilter port    no
  clag ISL           no                      clag ISL Oper UP     no
  clag role          unknown                 clag dual conn mac   0:0:0:0:0:0
  clag remote portID F.FFF                   clag system mac      0:0:0:0:0:0

The only way to recover a port that has been placed in the disabled state is to manually bring up the port with the sudo ifup <interface> command. See Interface Configuration and Management for more information about ifupdown.

Bringing up the disabled port does not correct the problem if the configuration on the connected end-station has not been resolved.

Bridge Assurance

On a point-to-point link where RSTP is running, if you want to detect unidirectional links and put the port in a discarding state (in error), you can enable bridge assurance on the port by enabling a port type network. The port is then in a bridge assurance inconsistent state until a BPDU is received from the peer. You need to configure the port type network on both ends of the link for bridge assurance to operate properly.

Bridge assurance is disabled by default.

To enable bridge assurance on an interface:

NCLU Commands

The following example commands enable bridge assurance on swp1:

cumulus@switch:~$ net add interface swp1 stp portnetwork
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the switch port interface stanza in the /etc/network/interfaces file and add the mstpctl-portnetwork yes line. The following example enables bridge assurance on swp5:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto swp5
iface swp5 
    mstpctl-portnetwork yes
...

Run ifreload -a to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To enable bridge assurance at runtime, run mstpctl:

cumulus@switch:~$ sudo mstpctl setportnetwork br1007 swp1.1007 yes

cumulus@switch:~$ sudo mstpctl showportdetail br1007 swp1.1007 | grep network
  network port       yes                     BA inconsistent      yes

To monitor logs for bridge assurance messages, run the following command:

cumulus@switch:~$ sudo grep -in assurance /var/log/syslog | grep mstp
  1365:Jun 25 18:03:17 mstpd: br1007:swp1.1007 Bridge assurance inconsistent

BPDU Filter

You can enable bpdufilter on a switch port, which filters BPDUs in both directions. This disables STP on the port as no BPDUs are transiting.

Using BDPU filter might cause layer 2 loops. Use this feature deliberately and with extreme caution.

To configure the BPDU filter on an interface:

NCLU Commands

The following example commands configure the BPDU filter on swp6:

cumulus@switch:~$ net add interface swp6 stp portbpdufilter
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the switch port interface stanza in the /etc/network/interfaces file and add the mstpctl-portbpdufilter yes line. The following example configures BPDU filter on swp6:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto swp6
iface swp6
    mstpctl-portbpdufilter yes
...

Run ifreload -a to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

Runtime Configuration (Advanced)

A runtime configuration is non-persistent, which means the configuration you create here does not persist after you reboot the switch.

To enable BPDU filter at runtime, run mstpctl. For example:

cumulus@switch:~$ sudo mstpctl setportbpdufilter br100 swp1.100=yes swp2.100=yes

Storm Control

Storm control provides protection against excessive inbound BUM (broadcast, unknown unicast, multicast) traffic on layer 2 switch port interfaces, which can cause poor network performance.

Storm control is not supported on a switch with the Tomahawk2 ASIC.

You configure storm control for each physical port by editing the /etc/cumulus/switchd.conf file.

For example, to enable broadcast storm control for swp1 at 400 packets per second (pps) and multicast storm control at 3000 pps, edit the /etc/cumulus/switchd.conf file and uncomment the storm_control.broadcast and storm_control.multicast lines:

cumulus@switch:~$ sudo nano /etc/cumulus/switchd.conf
...
# Storm Control setting on a port, in pps, 0 means disable
interface.swp1.storm_control.broadcast = 400
interface.swp1.storm_control.multicast = 3000
...

When you update the /etc/cumulus/switchd.conf file, you must restart switchd for the changes to take effect. Run the sudo systemctl restart switchd.service command.

Alternatively, you can run the following commands. The configuration below takes effect immediately, but does not persist if you reboot the switch. For a persistent configuration, edit the /etc/cumulus/switchd.conf file, as described above.

cumulus@switch:~$ sudo sh -c 'echo 400 > /cumulus/switchd/config/interface/swp1/storm_control/unknown_unicast'
cumulus@switch:~$ sudo sh -c 'echo 3000 > /cumulus/switchd/config/interface/swp1/storm_control/multicast'
cumulus@switch:~$ sudo systemctl restart switchd.service

Spanning Tree Parameter List

Spanning tree parameters are defined in the IEEE 802.1D and 802.1Q specifications.

The table below describes the STP configuration parameters available in Cumulus Linux. For a comparison of STP parameter configuration between mstpctl and other vendors, read this knowledge base article.

Most of these parameters are blacklisted in the ifupdown_blacklist section of the /etc/netd.conf file. Before you configure these parameters, you must edit the file to remove them from the blacklist.

Parameter
NCLU Command
Description
mstpctl-maxagenet add bridge stp maxage <seconds>Sets the maximum age of the bridge in seconds. The default is 20. The maximum age must meet the condition 2 * (Bridge Forward Delay - 1 second) >= Bridge Max Age.
mstpctl-ageingnet add bridge stp ageing <seconds>Sets the Ethernet (MAC) address ageing time for the bridge in seconds when the running version is STP, but not RSTP/MSTP. The default is 1800.
mstpctl-fdelaynet add bridge stp fdelay <seconds>Sets the bridge forward delay time in seconds. The default value is 15. The bridge forward delay must meet the condition 2 * (Bridge Forward Delay - 1 second) >= Bridge Max Age.
mstpctl-maxhopsnet add bridge stp maxhops <max-hops>Sets the maximum hops for the bridge. The default is 20.
mstpctl-txholdcountnet add bridge stp txholdcount <hold-count>Sets the bridge transmit hold count. The default value is 6.
mstpctl-forceversnet add bridge stp forcevers RSTP|STPSets the force STP version of the bridge to either RSTP/STP. MSTP is not supported currently. The default is RSTP.
mstpctl-treeprionet add bridge stp treeprio <priority>Sets the tree priority of the bridge for an MSTI (multiple spanning tree instance). The priority value is a number between 0 and 61440 and must be a multiple of 4096. The bridge with the lowest priority is elected the root bridge. The default is 32768.
Note: Cumulus Linux supports MSTI 0 only. It does not support MSTI 1 through 15.
mstpctl-hellonet add bridge stp hello <seconds>Sets the bridge hello time in seconds. The default is 2.
mstpctl-portpathcostnet add interface <interface> stp portpathcost <cost>Sets the port cost of the interface. The default is 0.
mstpd supports only long mode; 32 bits for the path cost.
mstpctl-treeportprionet add interface <interface> stp treeportprio <priority>Sets the priority of the interface for the MSTI. The priority value is a number between 0 and 240 and must be a multiple of 16. The default is 128.
Note: Cumulus Linux supports MSTI 0 only. It does not support MSTI 1 through 15.
mstpctl-portadminedgenet add interface <interface> stp portadminedgeEnables or disables the initial edge state of the interface in the bridge. The default is no.
In NCLU, to use a setting other than the default, you must specify this attribute without setting an option.
mstpctl-portautoedgenet add interface <interface> stp portautoedgeEnables or disables the auto transition to and from the edge state of the interface in the bridge. PortAutoEdge is enabled by default.
portautoedge is an enhancement to the standard PortAdminEdge (PortFast) mode, which allows for the automatic detection of edge ports.
Note: Edge ports and access ports are not the same thing. Edge ports transition directly to the forwarding state and skip the listening and learning stages. Upstream topology change notifications are not generated when an edge port’s link changes state. Access ports only forward untagged traffic; however, there is no such restriction on edge ports, which can forward both tagged and untagged traffic.
When a BPDU is received on a port configured with PortAutoEdge, the port ceases to be in the edge port state and transitions into a normal STP port.
When BPDUs are no longer received on the interface, the port becomes an edge port, and transitions through the discarding and learning states before resuming forwarding.
mstpctl-portp2pnet add interface <interface> stp portp2p yes|noEnables or disables the point-to-point detection mode of the interface in the bridge.
mstpctl-portrestrrolenet add interface <interface> stp portrestrroleEnables or disables the ability of the interface in the bridge to take the root role. The default is no.
To enable this feature with the NCLU command, you specify this attribute without an option (portrestrrole). To enable this feature by editing the /etc/network/interfaces file, you specify this attribute with yes (mstpctl-portrestrrole yes).
mstpctl-portrestrtcnnet add interface <interface> stp portrestrtcnEnables or disables the ability of the interface in the bridge to propagate received topology change notifications. The default is no.
mstpctl-portnetworknet add interface <interface> stp portnetworkEnables or disables the bridge assurance capability for a network interface. The default is no.
mstpctl-bpduguardnet add interface <interface> stp bpduguardEnables or disables the BPDU guard configuration of the interface in the bridge. The default is no. See above.
mstpctl-portbpdufilternet add interface <interface> stp portbpdufilterEnables or disables the BPDU filter functionality for an interface in the bridge. The default is no.
mstpctl-treeportcostnet add interface <interface> stp treeportcost <port-cost>Sets the spanning tree port cost to a value from 0 to 255. The default is 0.

Caveats and Errata

MSTP is not supported currently because Cumulus Linux only supports MSTI 0 (not MSTI 1 through 15). However, interoperability with MSTP networks can be accomplished using PVRSTP or PVSTP.

The source code for mstpd and mstpctl was written by Vitalii Demianets and is hosted at the URL below.

Link Layer Discovery Protocol

The lldpd daemon implements the IEEE802.1AB (Link Layer Discovery Protocol, or LLDP) standard. LLDP shows you which ports are neighbors of a given port. By default, lldpd runs as a daemon and starts at system boot. lldpd command line arguments are placed in /etc/default/lldpd. All lldpd configuration options are saved in /etc/lldpd.conf or under /etc/lldpd.d/.

For more details on the command line arguments and configuration options, see man lldpd(8).

lldpd supports CDP (Cisco Discovery Protocol, v1 and v2) and logs by default into /var/log/daemon.log with an lldpd prefix.

You can use the lldpcli CLI tool to query the lldpd daemon for neighbors, statistics, and other running configuration information. See man lldpcli(8) for details.

Configure LLDP

You configure lldpd settings in /etc/lldpd.conf or /etc/lldpd.d/.

Here is an example persistent configuration:

cumulus@switch:~$ sudo cat /etc/lldpd.conf
configure lldp tx-interval 40
configure lldp tx-hold 3
configure system interface pattern *,!eth0,swp*

The last line in the example above shows that LLDP is disabled on eth0. To disable LLDP on a single port, edit the /etc/default/lldpd file. This file specifies the default options to present to the lldpd service when it starts. The following example uses the -I option to disable LLDP on swp43:

cumulus@switch:~$ sudo nano /etc/default/lldpd

# Add "-x" to DAEMON_ARGS to start SNMP subagent
# Enable CDP by default
DAEMON_ARGS="-c -I *, !swp43"

lldpd has two timers defined by the tx-interval setting that affect each switch port:

lldpd logs to /var/log/daemon.log with the lldpd prefix:

cumulus@switch:~$ sudo tail -f /var/log/daemon.log  | grep lldp
Aug  7 17:26:17 switch lldpd[1712]: unable to get system name
Aug  7 17:26:17 switch lldpd[1712]: unable to get system name
Aug  7 17:26:17 switch lldpcli[1711]: lldpd should resume operations
Aug  7 17:26:32 switch lldpd[1805]: NET-SNMP version 5.4.3 AgentX subagent connected

Example lldpcli Commands

To show all neighbors on all ports and interfaces:

cumulus@switch:~$ sudo lldpcli show neighbors
-------------------------------------------------------------------------------
LLDP neighbors:
-------------------------------------------------------------------------------
Interface:    eth0, via: LLDP, RID: 1, Time: 0 day, 17:38:08
  Chassis:
    ChassisID:    mac 08:9e:01:e9:66:5a
    SysName:      PIONEERMS22
    SysDescr:     Cumulus Linux version 4.0.0 running on quanta lb9
    MgmtIP:       192.168.0.22
    Capability:   Bridge, on
    Capability:   Router, on
  Port:
    PortID:       ifname swp47
    PortDescr:    swp47
-------------------------------------------------------------------------------
Interface:    swp1, via: LLDP, RID: 10, Time: 0 day, 17:08:27
  Chassis:
    ChassisID:    mac 00:01:00:00:09:00
    SysName:      MSP-1
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.9
    MgmtIP:       fe80::201:ff:fe00:900
    Capability:   Bridge, off
    Capability:   Router, on
  Port:
    PortID:       ifname swp1
    PortDescr:    swp1
-------------------------------------------------------------------------------
Interface:    swp2, via: LLDP, RID: 10, Time: 0 day, 17:08:27
  Chassis:
    ChassisID:    mac 00:01:00:00:09:00
    SysName:      MSP-1
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.9
    MgmtIP:       fe80::201:ff:fe00:900
    Capability:   Bridge, off
    Capability:   Router, on
  Port:
    PortID:       ifname swp2
    PortDescr:    swp2
-------------------------------------------------------------------------------
Interface:    swp3, via: LLDP, RID: 11, Time: 0 day, 17:08:27
  Chassis:
    ChassisID:    mac 00:01:00:00:0a:00
    SysName:      MSP-2
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.10
    MgmtIP:       fe80::201:ff:fe00:a00
    Capability:   Bridge, off
    Capability:   Router, on
  Port:
    PortID:       ifname swp1
    PortDescr:    swp1
-------------------------------------------------------------------------------
Interface:    swp4, via: LLDP, RID: 11, Time: 0 day, 17:08:27
  Chassis:
    ChassisID:    mac 00:01:00:00:0a:00
    SysName:      MSP-2
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.10
    MgmtIP:       fe80::201:ff:fe00:a00
    Capability:   Bridge, off
    Capability:   Router, on
  Port:
    PortID:       ifname swp2
    PortDescr:    swp2
-------------------------------------------------------------------------------
Interface:    swp49s1, via: LLDP, RID: 9, Time: 0 day, 16:55:00
  Chassis:
    ChassisID:    mac 00:01:00:00:0c:00
    SysName:      TORC-1-2
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.12
    MgmtIP:       fe80::201:ff:fe00:c00
    Capability:   Bridge, on
    Capability:   Router, on
  Port:
    PortID:       ifname swp6
    PortDescr:    swp6
-------------------------------------------------------------------------------
Interface:    swp49s0, via: LLDP, RID: 9, Time: 0 day, 16:55:00
  Chassis:
    ChassisID:    mac 00:01:00:00:0c:00
    SysName:      TORC-1-2
    SysDescr:     Cumulus Linux version 4.0.0 running on QEMU Standard PC (i440FX + PIIX, 1996)
    MgmtIP:       192.0.2.12
    MgmtIP:       fe80::201:ff:fe00:c00
    Capability:   Bridge, on
    Capability:   Router, on
  Port:
    PortID:       ifname swp5
    PortDescr:    swp5
-------------------------------------------------------------------------------

To show lldpd statistics for all ports:

cumulus@switch:~$ sudo lldpcli show statistics
----------------------------------------------------------------------
LLDP statistics:
----------------------------------------------------------------------
Interface:    eth0
  Transmitted:  9423
  Received:     17634
  Discarded:    0
  Unrecognized: 0
  Ageout:       10
  Inserted:     20
  Deleted:      10
--------------------------------------------------------------------
Interface:    swp1
  Transmitted:  9423
  Received:     6264
  Discarded:    0
  Unrecognized: 0
  Ageout:       0
  Inserted:     2
  Deleted:      0
---------------------------------------------------------------------
Interface:    swp2
  Transmitted:  9423
  Received:     6264
  Discarded:    0
  Unrecognized: 0
  Ageout:       0
  Inserted:     2
  Deleted:      0
---------------------------------------------------------------------
Interface:    swp3
  Transmitted:  9423
  Received:     6265
  Discarded:    0
  Unrecognized: 0
  Ageout:       0
  Inserted:     2
  Deleted:      0
----------------------------------------------------------------------
...

To show lldpd statistics summary for all ports:

cumulus@switch:~$ sudo lldpcli show statistics summary
---------------------------------------------------------------------
LLDP Global statistics:
---------------------------------------------------------------------
Summary of stats:
  Transmitted:  648186
  Received:     437557
  Discarded:    0
  Unrecognized: 0
  Ageout:       10
  Inserted:     38
  Deleted:      10

To show the lldpd running configuration:

cumulus@switch:~$ sudo lldpcli show running-configuration
--------------------------------------------------------------------
Global configuration:
--------------------------------------------------------------------
Configuration:
  Transmit delay: 30
  Transmit hold: 4
  Receive mode: no
  Pattern for management addresses: (none)
  Interface pattern: (none)
  Interface pattern blacklist: (none)
  Interface pattern for chassis ID: (none)
  Override description with: (none)
  Override platform with: Linux
  Override system name with: (none)
  Advertise version: yes
  Update interface descriptions: no
  Promiscuous mode on managed interfaces: no
  Disable LLDP-MED inventory: yes
  LLDP-MED fast start mechanism: yes
  LLDP-MED fast start interval: 1
  Source MAC for LLDP frames on bond slaves: local
  Portid TLV Subtype for lldp frames: ifname
--------------------------------------------------------------------
Runtime Configuration (Advanced)

A runtime configuration does not persist when you reboot the switch; all changes are lost.

To configure active interfaces:

cumulus@switch:~$ sudo lldpcli configure system interface pattern "swp*"

To configure inactive interfaces:

cumulus@switch:~$ sudo lldpcli configure system interface pattern *,!eth0,swp*

The active interface list always overrides the inactive interface list.

To reset any interface list to none:

cumulus@switch:~$ sudo lldpcli configure system interface pattern ""

Enable the SNMP Subagent in LLDP

LLDP does not enable the SNMP subagent by default. You need to edit /etc/default/lldpd and enable the -x option.

cumulus@switch:~$ sudo nano /etc/default/lldpd

# Add "-x" to DAEMON_ARGS to start SNMP subagent

# Enable CDP by default
DAEMON_ARGS="-c"

Caveats and Errata

Annex E (and hence Annex D) of IEEE802.1AB (lldp) is not supported.

Voice VLAN

In Cumulus Linux, a voice VLAN is a VLAN dedicated to voice traffic on a switch port. However, the term can mean different things to different vendors.

Voice VLAN is part of a trunk port with two VLANs that comprises either of the following:

The voice traffic is an 802.1q-tagged packet with a VLAN ID (that might or might not be 0) and an 802.1p (3-bit layer 2 COS) with a specific value (typically 5 is assigned for voice traffic).

Data traffic is always untagged.

Voice VLAN Configuration Example

In this example configuration:

To configure the topology shown above:

NCLU Commands
cumulus@switch:~$ net add bridge bridge ports swp1-3
cumulus@switch:~$ net add bridge bridge vids 1-1000
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net add interface swp1 bridge voice-vlan 200
cumulus@switch:~$ net add interface swp2 bridge voice-vlan 100 data-vlan 10
cumulus@switch:~$ net add interface swp3 bridge voice-vlan 300
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add the following configuration:

cumulus@switch:~$ sudo nano /etc/network/interfaces

auto swp1
iface swp1
    bridge-vids 200
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto swp2
iface swp2
    bridge-pvid 10
    bridge-vids 100
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto swp3
iface swp3
    bridge-vids 300
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto bridge
iface bridge
  bridge-ports swp1 swp2 swp3
  bridge-pvid 1
  bridge-vids 1-1000
  bridge-vlan-aware yes

Configure LLDP

Configuring voice VLAN with NCLU does not configure lldpd in Cumulus Linux; therefore, LLDP-MED does not provide data and voice VLAN information. You can configure LLDP-MED for each interface in a new file in /etc/lldp.d. In the following example, the file is called /etc/lldpd.d/voice_vlan.conf:

cumulus@switch:~$ sudo nano /etc/lldpd.d/voice_vlan.conf
configure ports swp1 med policy application voice tagged vlan 200 priority voice dscp 46
configure ports swp2 med policy application voice tagged vlan 100 priority voice dscp 46
configure ports swp3 med policy application voice tagged vlan 300 priority voice dscp 46

You can also use the lldpcli command to configure an LLDP-MED network policy. However, lldpcli commands do not persist across switch reboots.

Troubleshooting

To show the bridge-vids:

NCLU Commands

Run the net show bridge vlan command:

cumulus@switch:~$ net show bridge vlan

Interface      VLAN  Flags
-----------  ------  ---------------------
swp1            1  PVID, Egress Untagged
              200

swp2           10  PVID, Egress Untagged
              200

swp3            1  PVID, Egress Untagged
              300
Linux Commands

Run the bridge fdb show command:

cumulus@switch:~$ bridge fdb show
44:38:39:00:12:9c dev swp2 VLAN 0 master bridge-A permanent
44:38:39:00:12:9b dev swp1 VLAN 0 master bridge-A permanent
44:38:39:00:12:9e dev swp4 VLAN 0 master bridge-B permanent
44:38:39:00:12:9d dev swp3 VLAN 0 master bridge-B permanent

To obtain MAC address information:

NCLU Commands

Run the net show bridge macs command:

cumulus@switch:~$ net show bridge macs

VLAN      Master    Interface    MAC                   TunnelDest  State      Flags    LastSeen
--------  --------  -----------  -----------------  -------------  ---------  -------  ----------
untagged  bridge    bridge       08:00:27:d5:00:93                 permanent           00:13:54
untagged  bridge    swp1         08:00:27:6a:ad:da                 permanent           00:13:54
untagged  bridge    swp2         08:00:27:e3:0c:a7                 permanent           00:13:54
untagged  bridge    swp3         08:00:27:9e:98:86                 permanent           00:13:54
Linux Commands

Run the sudo brctl showmacs <bridge> command:

cumulus@switch:~$ sudo brctl showmacs my_bridge
  port name mac addr              is local?       ageing timer
  swp4      06:90:70:22:a6:2e     no                19.47
  swp1      12:12:36:43:6f:9d     no                40.50
  bond0     2a:95:22:94:d1:f0     no                 1.98
  swp1      44:38:39:00:12:9b     yes                0.00
  swp2      44:38:39:00:12:9c     yes                0.00
  swp3      44:38:39:00:12:9d     yes                0.00
  swp4      44:38:39:00:12:9e     yes                0.00
  bond0     44:38:39:00:12:9f     yes                0.00
  swp2      90:e2:ba:2c:b1:94     no                12.84
  swp2      a2:84:fe:fc:bf:cd     no                 9.43

To capture LLDP information, check syslog or use tcpdump on an interface.

Caveats and Errata

Bonding - Link Aggregation

Linux bonding provides a method for aggregating multiple network interfaces (slaves) into a single logical bonded interface (bond). Link aggregation is useful for linear scaling of bandwidth, load balancing, and failover protection.

Cumulus Linux supports two bonding modes:

Cumulus Linux uses version 1 of the LAG control protocol (LACP).

To temporarily bring up a bond even when there is no LACP partner, use LACP Bypass.

Hash Distribution

Egress traffic through a bond is distributed to a slave based on a packet hash calculation, providing load balancing over the slaves; many conversation flows are distributed over all available slaves to load balance the total traffic. Traffic for a single conversation flow always hashes to the same slave.

The hash calculation uses packet header data to choose to which slave to transmit the packet:

In a failover event, the hash calculation is adjusted to steer traffic over available slaves.

Create a Bond

In the example below, the front panel port interfaces swp1 thru swp4 are slaves in bond0, while swp5 and swp6 are not part of bond0.

To create and configure a bond:

NCLU Commands

Run the net add bond command. The example command below creates a bond called bond0 with slaves swp1, swp2, swp3, and swp4:

cumulus@switch:~$ net add bond bond0 bond slaves swp1-4
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add a stanza for the bond. The example below creates a bond called bond0 with slaves swp1, swp2, swp3, and swp4:

cumulus@switch:~$ sudo nanno /etc/network/interfaces
...
auto bond0
iface bond0
    bond-slaves swp1 swp2 swp3 swp4
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

  • The bond is configured by default in IEEE 802.3ad link aggregation mode. To configure the bond in balance-xor mode, see Configuration Parameters below.
  • If the bond is not going to become part of a bridge, you need to specify an IP address.
  • The name of the bond must be compliant with Linux interface naming conventions and unique within the switch.

When networking is started on the switch, bond0 is created as MASTER and interfaces swp1 thru swp4 come up in SLAVE mode, as seen in the ip link show command:

cumulus@switch:~$ ip link show
...

3: swp1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
4: swp2: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
5: swp3: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
6: swp4: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond0 state UP mode DEFAULT qlen 500
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff
...

55: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT
    link/ether 44:38:39:00:03:c1 brd ff:ff:ff:ff:ff:ff

All slave interfaces within a bond have the same MAC address as the bond. Typically, the first slave added to the bond donates its MAC address as the bond MAC address, whereas the MAC addresses of the other slaves are set to the bond MAC address. The bond MAC address is used as the source MAC address for all traffic leaving the bond and provides a single destination MAC address to address traffic to the bond.

Configure Bond Options

The configuration options for a bond are are described in the table below. To configure a bond:

NCLU Commands

Run net add bond <bond-name> bond <option>. The following example sets the bond mode for bond01 to balance-xor:

cumulus@switch:~$ net add bond bond1 bond mode balance-xor
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add the parameter to the bond stanza, then load the new configuration. The following example sets the bond mode for bond01 to balance-xor:

cumulus@switch:~$ sudo nanno /etc/network/interfaces
...
auto bond1
iface bond1
    bond-mode balance-xor
    bond-slaves swp1 swp2 swp3 swp4
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

Each bond configuration option, except for bond slaves, is set to the recommended value by default in Cumulus Linux. Only configure an option if a different setting is needed. For more information on configuration values, refer to the Related Information section below.

ParameterDescription
bond-mode 802.3ad|balance-xorCumulus Linux supports IEEE 802.3ad link aggregation mode (802.3ad) and balance-xor mode.
The default mode is 802.3ad.

Note: When you enable balance-xor mode, the bonding of slave interfaces are static and all slave interfaces are active for load balancing and fault tolerance purposes. Packet transmission on the bond is based on the hash policy specified by xmit-hash-policy.

When using balance-xor mode to dual-connect host-facing bonds in an MLAG environment, you must configure the clag-id parameter on the MLAG bonds and it must be the same on both MLAG switches. Otherwise, the bonds are treated by the MLAG switch pair as single-connected.

Use balance-xor mode only if you cannot use LACP; LACP can detect mismatched link attributes between bond members and can even detect misconnections.
bond-slaves <interface-list>The list of slaves in the bond.
bond miimon <value>Defines how often the link state of each slave is inspected for failures. You can specify a value between 0 and 255. The default value is 100.
bond downdelay <milliseconds>Specifies the time, in milliseconds (between 0 and 65535), to wait before disabling a slave after a link failure is detected. The default value is 0.

This option is only valid for the miimon link monitor. The downdelay value must be a multiple of the miimon value; if not, it is rounded down to the nearest multiple.
bond-updelay <milliseconds>Specifies the time, in milliseconds (between 0 and 65535), to wait before enabling a slave after a link recovery is detected. The default value is 0.

This option is only valid for the miimon link monitor. The updelay value must be a multiple of the miimon value; if not, it is rounded down to the nearest multiple.
bond-use-carrier noDetermines the link state.
bond-lacp-bypass-allowEnables LACP bypass.
bond-lacp-rate slowSets the rate to ask the link partner to transmit LACP control packets. slow is the only option.
bond-min-linksDefines the minimum number of links (between 0 and 255) that must be active before the bond is put into service. The default value is 1.

A value greater than 1 is useful if higher level services need to ensure a minimum aggregate bandwidth level before activating a bond. Keeping bond-min-links set to 1 indicates the bond must have at least one active member. If the number of active members drops below the bond-min-links setting, the bond appears to upper-level protocols as link-down. When the number of active links returns to greater than or equal to bond-min-links, the bond becomes link-up.

Show Bond Information

To show information for a bond:

NCLU Commands

Run the net show interface <bond> command:

cumulus@switch:~$ net show interface bond1
    Name    MAC                Speed    MTU    Mode
--  ------  -----------------  -------  -----  ------
UP  bond1   00:02:00:00:00:12  20G      1500   Bond


Bond Details
---------------  -------------
Bond Mode:       Balance-XOR
Load Balancing:  Layer3+4
Minimum Links:   1
In CLAG:         CLAG Inactive


    Port     Speed      TX    RX    Err    Link Failures
--  -------  -------  ----  ----  -----  ---------------
UP  swp3(P)  10G         0     0      0                0
UP  swp4(P)  10G         0     0      0                0


LLDP
-------  ----  ------------
swp3(P)  ====  swp1(p1c1h1)
swp4(P)  ====  swp2(p1c1h1)Routing
-------
  Interface bond1 is up, line protocol is up
  Link ups:       3    last: 2017/04/26 21:00:38.26
  Link downs:     2    last: 2017/04/26 20:59:56.78
  PTM status: disabled
  vrf: Default-IP-Routing-Table
  index 31 metric 0 mtu 1500
  flags: <UP,BROADCAST,RUNNING,MULTICAST>
  Type: Ethernet
  HWaddr: 00:02:00:00:00:12
  inet6 fe80::202:ff:fe00:12/64
  Interface Type Other
Linux Commands

Run the sudo cat /proc/net/bonding/<bond> command:

cumulus@switch:~$ sudo cat /proc/net/bonding/bond01

Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: load balancing (xor)
Transmit Hash Policy: layer3+4 (1)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0


Slave Interface: swp1
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:00:03
Slave queue ID: 0

The detailed output in /proc/net/bonding/<filename> includes the actor/partner LACP information. This information is not necessary and requires you to use sudo to view the file.

Caveats and Errata

swp49 xe0 0 0 -1 0
swp50 xe1 0 0 -1 0
swp51 xe1 1 0 -1 0
swp52 xe0 1 0 -1 0

Single port member bonds, bonds with different units (xe0 or xe1, as above), or layer 3 bonds do not have this issue.

On Cumulus RMP switches, which are built with two Hurricane2 ASICs, you cannot form an LACP bond on links that terminate on different Hurricane2 ASICs.

Ethernet Bridging - VLANs

Ethernet bridges enable hosts to communicate through layer 2 by connecting all of the physical and logical interfaces in the system into a single layer 2 domain. The bridge is a logical interface with a MAC address and an MTU (maximum transmission unit). The bridge MTU is the minimum MTU among all its members. By default, the bridge's MAC address is the MAC address of the first port in the bridge-ports list. The bridge can also be assigned an IP address, as discussed below.

Bridge members can be individual physical interfaces, bonds, or logical interfaces that traverse an 802.1Q VLAN trunk.

Cumulus Networks recommends using VLAN-aware mode bridges instead of traditional mode bridges. The bridge driver in Cumulus Linux is capable of VLAN filtering, which allows for configurations that are similar to incumbent network devices. For a comparison of traditional and VLAN-aware modes, read this knowledge base article.

  • Cumulus Linux does not put all ports into a bridge by default.
  • You can configure both VLAN-aware and traditional mode bridges on the same network in Cumulus Linux; however you cannot have more than one VLAN-aware bridge on a given switch.

Create a VLAN-aware Bridge

To create a VLAN-aware bridge, see VLAN-aware Bridge Mode.

Create a Traditional Mode Bridge

To create a traditional mode bridge, see Traditional Bridge Mode.

Bridge MAC Addresses

The MAC address for a frame is learned when the frame enters the bridge through an interface. The MAC address is recorded in the bridge table and the bridge forwards the frame to its intended destination by looking up the destination MAC address. The MAC entry is then maintained for a period of time defined by the bridge-ageing configuration option. If the frame is seen with the same source MAC address before the MAC entry age is exceeded, the MAC entry age is refreshed; if the MAC entry age is exceeded, the MAC address is deleted from the bridge table.

The following example output shows a MAC address table for the bridge:

cumulus@switch:~$ net show bridge macs 
VLAN      Master    Interface    MAC                  TunnelDest  State      Flags    LastSeen
--------  --------  -----------  -----------------  ------------  ---------  -------  -----------------
untagged  bridge    swp1         44:38:39:00:00:03                                    00:00:15
untagged  bridge    swp1         44:38:39:00:00:04                permanent           20 days, 01:14:03

By default, Cumulus Linux stores MAC addresses in the Ethernet switching table for 1800 seconds (30 minutes). To change the amount of time MAC addresses are stored in the table, configure bridge ageing.

The bridge ageing option is in the NCLU blacklist. If you want to change this setting, you need to first remove the bridge-ageing keyword from the ifupdown_blacklist section of the /etc/netd.conf file, then restart the netd service.

To configure bridge ageing:

NCLU Commands

Run the net add bridge bridge ageing command. The following example commands set MAC address ageing to 600 seconds:

cumulus@switch:~$ net add bridge bridge ageing 600
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add bridge-ageing to the bridge stanza. The following example sets MAC address ageing to 600 seconds.

cumulus@switch:~$ sudo nano /etc/network/interfaces 
...
auto bridge
iface bridge
    bridge-ageing 600
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

Configure an SVI (Switch VLAN Interface)

Bridges can be included as part of a routing topology after being assigned an IP address. This enables hosts within the bridge to communicate with other hosts outside of the bridge through a switch VLAN interface (SVI), which provides layer 3 routing. The IP address of the bridge is typically from the same subnet as the member hosts of the bridge.

When you add an interface to a bridge, it ceases to function as a router interface and the IP address on the interface becomes unreachable.

To configure the SVI:

NCLU Commands

Run the net add bridge and net add vlan commands. The following example commands configure an SVI using swp1 and swp2, and VLAN ID 10.

cumulus@switch:~$ net add bridge bridge ports swp1-2
cumulus@switch:~$ net add vlan 10 ip address 10.100.100.1/24
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file to add the interfaces and VLAN ID you want to use. The following configures an SVI using swp1 and swp2, and VLAN ID 10. The bridge-vlan-aware parameter associates the SVI with the VLAN-aware bridge.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-vids 10
    bridge-vlan-aware yes

auto bridge.10
iface bridge.10
    address 10.100.100.1/24
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

When you configure a switch initially, all southbound bridge ports might be down; therefore, by default, the SVI is also down. You can force the SVI to always be up by disabling interface state tracking, which leaves the SVI in the UP state always, even if all member ports are down. Other implementations describe this feature as no autostate. This is beneficial if you want to perform connectivity testing.

To keep the SVI perpetually UP, create a dummy interface, then make the dummy interface a member of the bridge.

Example Configuration

Consider the following configuration, without a dummy interface in the bridge:

cumulus@switch:~$ sudo cat /etc/network/interfaces
...

auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge-ports swp3
    bridge-vids 100
    bridge-pvid 1
...

With this configuration, when swp3 is down, the SVI is also down:

cumulus@switch:~$ ip link show swp3
5: swp3: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast master bridge state DOWN mode DEFAULT group default qlen 1000
    link/ether 2c:60:0c:66:b1:7f brd ff:ff:ff:ff:ff:ff
cumulus@switch:~$ ip link show bridge
35: bridge: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN mode DEFAULT group default
    link/ether 2c:60:0c:66:b1:7f brd ff:ff:ff:ff:ff:ff

Now add the dummy interface to your network configuration:

  1. Edit the /etc/network/interfaces file and add the dummy interface stanza before the bridge stanza:
cumulus@switch:~$ sudo nano /etc/network/interfaces
...

auto dummy
iface dummy
    link-type dummy

auto bridge
iface bridge
...
  1. Add the dummy interface to the bridge-ports line in the bridge configuration:
auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge-ports swp3 dummy
    bridge-vids 100
    bridge-pvid 1
  1. Save and exit the file, then reload the configuration:
cumulus@switch:~$ sudo ifreload -a

Now, even when swp3 is down, both the dummy interface and the bridge remain up:

cumulus@switch:~$ ip link show swp3
5: swp3: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast master bridge state DOWN mode DEFAULT group default qlen 1000
    link/ether 2c:60:0c:66:b1:7f brd ff:ff:ff:ff:ff:ff
cumulus@switch:~$ ip link show dummy
37: dummy: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue master bridge state UNKNOWN mode DEFAULT group default
    link/ether 66:dc:92:d4:f3:68 brd ff:ff:ff:ff:ff:ff
cumulus@switch:~$ ip link show bridge
35: bridge: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default
    link/ether 2c:60:0c:66:b1:7f brd ff:ff:ff:ff:ff:ff

By default, Cumulus Linux automatically generates IPv6 link-local addresses on VLAN interfaces. If you want to use a different mechanism to assign link-local addresses, you can disable this feature. You can disable link-local automatic address generation for both regular IPv6 addresses and address-virtual (macvlan) addresses.

To disable automatic address generation for a regular IPv6 address on a VLAN:

NCLU Commands

Run the net add vlan <vlan> ipv6-addrgen off command. The following example command disables automatic address generation for a regular IPv6 address on a VLAN 100.

cumulus@switch:~$ net add vlan 100 ipv6-addrgen off
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add the line ipv6-addrgen off to the VLAN stanza. The following example disables automatic address generation for a regular IPv6 address on VLAN 100.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto vlan100
iface vlan 100
    ipv6-addrgen off
    vlan-id 100
    vlan-raw-device bridge
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

To re-enable automatic link-local address generation for a VLAN:

NCLU Commands

Run the net del vlan <vlan> ipv6-addrgen off command. The following example command re-enables automatic address generation for a regular IPv6 address on VLAN 100.

cumulus@switch:~$ net del vlan 100 ipv6-addrgen off
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands
  1. Edit the /etc/network/interfaces file and remove the line ipv6-addrgen off from the VLAN stanza. The following example re-enables automatic address generation for a regular IPv6 address on a VLAN 100.
  2. Run the ifreload -a command to load the new configuration:
cumulus@switch:~$ ifreload -a

bridge fdb Command Output

The bridge fdb command in Linux interacts with the forwarding database table (FDB), which the bridge uses to store the MAC addresses it learns and the ports on which it learns those MAC addresses. The bridge fdb show command output contains some specific keywords:

KeywordDescription
selfThe Linux kernel FDB entry flag that indicates the FDB entry belongs to the FDB on the device referenced by the device.
For example, this FDB entry belongs to the VXLAN device vx-1000: 00:02:00:00:00:08 dev vx-1000 dst 27.0.0.10 self
masterThe Linux kernel FDB entry flag that indicates the FDB entry belongs to the FDB on the device’s master and the FDB entry is pointing to a master’s port.
For example, this FDB entry is from the master device named bridge and is pointing to the VXLAN bridge port vx-1001: 02:02:00:00:00:08 dev vx-1001 vlan 1001 master bridge
offloadThe Linux kernel FDB entry flag that indicates the FDB entry is managed (or offloaded) by an external control plane, such as the BGP control plane for EVPN.

The following example shows the bridge fdb show command output:

cumulus@switch:~$ bridge fdb show | grep 02:02:00:00:00:08
02:02:00:00:00:08 dev vx-1001 vlan 1001 offload master bridge 
02:02:00:00:00:08 dev vx-1001 dst 27.0.0.10 self offload

  • 02:02:00:00:00:08 is the MAC address learned with BGP EVPN.
  • The first FDB entry points to a Linux bridge entry that points to the VXLAN device vx-1001.
  • The second FDB entry points to the same entry on the VXLAN device and includes additional remote destination information.
  • The VXLAN FDB augments the bridge FDB with additional remote destination information.
  • All FDB entries that point to a VXLAN port appear as two entries. The second entry augments the remote destination information.

Caveats and Errata

VLAN-aware Bridge Mode

The Cumulus Linux bridge driver supports two configuration modes, one that is VLAN-aware, and one that follows a more traditional Linux bridge model.

For traditional Linux bridges, the kernel supports VLANs in the form of VLAN subinterfaces. Enabling bridging on multiple VLANs means configuring a bridge for each VLAN and, for each member port on a bridge, creating one or more VLAN subinterfaces out of that port. This mode poses scalability challenges in terms of configuration size as well as boot time and run time state management, when the number of ports times the number of VLANs becomes large.

The VLAN-aware mode in Cumulus Linux implements a configuration model for large-scale layer 2 environments, with one single instance of spanning tree protocol. Each physical bridge member port is configured with the list of allowed VLANs as well as its port VLAN ID, either primary VLAN Identifier (PVID) or native VLAN. MAC address learning, filtering and forwarding are VLAN-aware. This significantly reduces the configuration size, and eliminates the large overhead of managing the port/VLAN instances as subinterfaces, replacing them with lightweight VLAN bitmaps and state updates.

You can configure both VLAN-aware and traditional mode bridges on thesame network in Cumulus Linux; however do not have more than one VLAN-aware bridge on a given switch.

Configure a VLAN-aware Bridge

The example below shows the commands required to create a VLAN-aware bridge configured for STP that contains two switch ports and includes 3 VLANs; the tagged VLANs 100 and 200 and the untagged (native) VLAN of 1.

NCLU Commands
cumulus@switch:~$ net add bridge bridge ports swp1-2
cumulus@switch:~$ net add bridge bridge vids 100,200
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The above commands create the following code snippet in the /etc/network/interfaces file:

auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 100 200
    bridge-vlan-aware yes
Linux Commands

Edit the /etc/network/interfaces file and add the bridge. An example configuration is shown below.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-vids 100 200
    bridge-pvid 1
    bridge-vlan-aware yes
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

The Primary VLAN Identifer (PVID) of the bridge defaults to 1. You do not have to specify bridge-pvid for a bridge or a port. However, even though this does not affect the configuration, it helps other users for readability. The following configurations are identical to each other and the configuration above:

auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-vids 1 100 200
    bridge-vlan-aware yes
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 1 100 200
    bridge-vlan-aware yes
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-vids 100 200
    bridge-vlan-aware yes

If you specify bridge-vids or bridge-pvid at the bridge level, these configurations are inherited by all ports in the bridge. However, specifying any of these settings for a specific port overrides the setting in the bridge.

Do not try to bridge the management port, eth0, with any switch ports (swp0, swp1 and so on). For example, if you create a bridge with eth0 and swp1, it will not work properly and might disrupt access to the management interface.

Reserved VLAN Range

For hardware data plane internal operations, the switching silicon requires VLANs for every physical port, Linux bridge, and layer 3 subinterface. Cumulus Linux reserves a range of 1000 VLANs by default; the reserved range is 3000-3999.

You can modify the reserved range if it conflicts with any user-defined VLANs, as long the new range is a contiguous set of VLANs with IDs anywhere between 2 and 4094, and the minimum size of the range is 300 VLANs.

To configure the reserved range:

  1. Open /etc/cumulus/switchd.conf in a text editor.
  2. Uncomment the following line, specify a new range, and save the file:
resv_vlan_range
  1. Restart switchd to implement the change:
cumulus@switch:~$ sudo systemctl restart switchd.service

While restarting switchd, all running ports will flap, and forwarding will be interrupted.

VLAN Filtering (VLAN Pruning)

By default, the bridge port inherits the bridge VIDs. To configure a port to override the bridge VIDs:

NCLU Commands

The following example commands configure swp3 to override the bridge VIDs:

cumulus@switch:~$ net add bridge bridge ports swp1-3
cumulus@switch:~$ net add bridge bridge vids 100,200
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net add interface swp3 bridge vids 200
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The above commands create the following code snippets in the /etc/network/interfaces file:

auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3
    bridge-pvid 1
    bridge-vids 100 200
    bridge-vlan-aware yes

auto swp3
iface swp3
  bridge-vids 200
Linux Commands

Edit the /etc/network/interfaces file. The following example commands configure swp3 to override the bridge VIDs:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2 swp3
    bridge-pvid 1
    bridge-vids 100 200
    bridge-vlan-aware yes

auto swp3
iface swp3
  bridge-vids 200
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

Untagged/Access Ports

Access ports ignore all tagged packets. In the configuration below, swp1 and swp2 are configured as access ports, while all untagged traffic goes to VLAN 100:

NCLU Commands
cumulus@switch:~$ net add bridge bridge ports swp1-2
cumulus@switch:~$ net add bridge bridge vids 100,200
cumulus@switch:~$ net add bridge bridge pvid 1
cumulus@switch:~$ net add interface swp1 bridge access 100
cumulus@switch:~$ net add interface swp2 bridge access 100
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

The above commands create the following code snippets in the /etc/network/interfaces file:

auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 100 200
    bridge-vlan-aware yes

auto swp1
iface swp1
    bridge-access 100

auto swp2
iface swp2
    bridge-access 100
Linux Commands

Edit the /etc/network/interfaces file.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 100 200
    bridge-vlan-aware yes

auto swp1
iface swp1
    bridge-access 100

auto swp2
iface swp2
    bridge-access 100
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

Drop Untagged Frames

With VLAN-aware bridge mode, you can configure a switch port to drop any untagged frames. To do this, add bridge-allow-untagged no to the switch port (not to the bridge). This leaves the bridge port without a PVID and drops untagged packets.

NCLU Commands

To configure a switch port to drop untagged frames, run the net add interface swp2 bridge allow-untagged no command. The following example command configures swp2 to drop untagged frames:

cumulus@switch:~$ net add interface swp2 bridge allow-untagged no

When you check VLAN membership for that port, it shows that there is no untagged VLAN.

cumulus@switch:~$ net show bridge vlan

Interface      VLAN   Flags
-----------  ------   ---------------------
swp1              1   PVID, Egress Untagged
                 10
                100
                200
swp2             10
                100
                200
Linux Commands

Edit the /etc/network/interfaces file to add the bridge-allow-untagged no line to the under the switch port interface stanza. The following example configures swp2 to drop untagged frames:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto swp1
iface swp1

auto swp2
iface swp2
    bridge-allow-untagged no 

auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 10 100 200
    bridge-vlan-aware yes
...

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ sudo ifreload -a

When you check VLAN membership for that port, it shows that there is no untagged VLAN.

cumulus@switch:~$ bridge -c vlan show
portvlan ids
swp1 1 PVID Egress Untagged
  10 100 200

swp2 10 100 200

bridge 1

VLAN Layer 3 Addressing - Switch Virtual Interfaces and Other VLAN Attributes

When configuring the VLAN attributes for the bridge, specify the attributes for each VLAN interface. If you are configuring the SVI for the native VLAN, you must declare the native VLAN and specify its IP address. Specifying the IP address in the bridge stanza itself returns an error.

NCLU Commands

The following example commands declare native VLAN 100 with IPv4 address 192.168.10.1/24 and IPv6 address 2001:db8::1/32.

cumulus@switch:~$ net add vlan 100 ip address 192.168.10.1/24
cumulus@switch:~$ net add vlan 100 ipv6 address 2001:db8::1/32
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file, then reload the configuration with the ifreload -a command. The following example declares native VLAN 100 with IPv4 address 192.168.10.1/24 and IPv6 address 2001:db8::1/32.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 swp2
    bridge-pvid 1
    bridge-vids 10 100 200
    bridge-vlan-aware yes

auto vlan100
iface vlan100
    address 192.168.10.1/24
    address 2001:db8::1/32
    vlan-id 100
    vlan-raw-device bridge
...

In the above configuration, if your switch is configured for multicast routing, you do not need to specify bridge-igmp-querier-src, as there is no need for a static IGMP querier configuration on the switch. Otherwise, the static IGMP querier configuration helps to probe the hosts to refresh their IGMP reports.

Configure ARP Timers

Cumulus Linux does not often interact directly with end systems as much as end systems interact with one another. Therefore, after a successful address resolution protocol (ARP) places a neighbor into a reachable state, Cumulus Linux might not interact with the client again for a long enough period of time for the neighbor to move into a stale state. To keep neighbors in the reachable state, Cumulus Linux includes a background process (/usr/bin/neighmgrd). The background process tracks neighbors that move into a stale, delay, or probe state, and attempts to refresh their state before they are removed from the Linux kernel and from hardware forwarding. The neighmgrd process only adds a neighbor if the sender’s IP in the ARP packet is in one of the SVI’s subnets (you can disable this check by setting subnet_checks to 0 in the /etc/cumulus/neighmgr.conf file).

The ARP refresh timer defaults to 1080 seconds (18 minutes). To change this setting, follow the procedures outlined in this knowledge base article.

Configure Multiple Ports in a Range

To save time, you can specify a range of ports or VLANs instead of enumerating each one individually.

To specify a range:

NCLU Commands

In the example below, swp1-52 indicates that swp1 through swp52 are part of the bridge.

cumulus@switch:~$ net add bridge bridge ports swp1-52
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

The glob keyword referenced in the bridge-ports attribute indicates that swp1 through swp52 are part of the bridge:

...
auto bridge
iface bridge
        bridge-vlan-aware yes
        bridge-ports glob swp1-52
        bridge-stp on
        bridge-vids 310 700 707 712 850 910
...

Example Configurations

The following sections provide example VLAN-aware bridge configurations.

Access Ports and Pruned VLANs

The following example configuration contains an access port and switch port that are pruned; they only sends and receive traffic tagged to and from a specific set of VLANs declared by the bridge-vids attribute. It also contains other switch ports that send and receive traffic from all the defined VLANs.

...
# ports swp3-swp48 are trunk ports which inherit vlans from the 'bridge'
# ie vlans 310,700,707,712,850,910
#
auto bridge
iface bridge
      bridge-ports swp1 swp2 swp3 ... swp51 swp52
      bridge-vids 310 700 707 712 850 910
      bridge-vlan-aware yes

auto swp1
iface swp1
      bridge-access 310
      mstpctl-bpduguard yes
      mstpctl-portadminedge yes

# The following is a trunk port that is "pruned".
# native vlan is 1, but only .1q tags of 707, 712, 850 are
# sent and received
#
auto swp2
iface swp2
      mstpctl-bpduguard yes
      mstpctl-portadminedge yes
      bridge-vids 707 712 850

# The following port is the trunk uplink and inherits all vlans
# from 'bridge'; bridge assurance is enabled using 'portnetwork' attribute
auto swp49
iface swp49
      mstpctl-portnetwork yes
      mstpctl-portpathcost 10

# The following port is the trunk uplink and inherits all vlans
# from 'bridge'; bridge assurance is enabled using 'portnetwork' attribute
 auto swp50
 iface swp50
      mstpctl-portnetwork yes
      mstpctl-portpathcost 0
...

Large Bond Set Configuration

The configuration below demonstrates a VLAN-aware bridge with a large set of bonds. The bond configurations are generated from a Mako template.

...
#
# vlan-aware bridge with bonds example
#
# uplink1, peerlink and downlink are bond interfaces.
# 'bridge' is a vlan aware bridge with ports uplink1, peerlink
# and downlink (swp2-20).
#
# native vlan is by default 1
#
# 'bridge-vids' attribute is used to declare vlans.
# 'bridge-pvid' attribute is used to specify native vlans if other than 1
# 'bridge-access' attribute is used to declare access port
#
auto lo
iface lo

auto eth0
iface eth0 inet dhcp

# bond interface
auto uplink1
iface uplink1
    bond-slaves swp32
    bridge-vids 2000-2079

# bond interface
auto peerlink
iface peerlink
    bond-slaves swp30 swp31
    bridge-vids 2000-2079 4094

# bond interface
auto downlink
iface downlink
    bond-slaves swp1
    bridge-vids 2000-2079

#
# Declare vlans for all swp ports
# swp2-20 get vlans from 2004 to 2022.
# The below uses mako templates to generate iface sections
# with vlans for swp ports
#
%for port, vlanid in zip(range(2, 20), range(2004, 2022)) :
    auto swp${port}
    iface swp${port}
      bridge-vids ${vlanid}

%endfor

# svi vlan 2000
auto bridge.2000
iface bridge.2000
    address 11.100.1.252/24

# l2 attributes for vlan 2000
auto bridge.2000
vlan bridge.2000
    bridge-igmp-querier-src 172.16.101.1

#
# vlan-aware bridge
#
auto bridge
iface bridge
    bridge-ports uplink1 peerlink downlink swp1 swp2 swp49 swp50
    bridge-vlan-aware yes

# svi peerlink vlan
auto peerlink.4094
iface peerlink.4094
    address 192.168.10.1/30
    broadcast 192.168.10.3
...

VXLANs with VLAN-aware Bridges

Cumulus Linux supports using VXLANs with VLAN-aware bridge configuration. This provides improved scalability, as multiple VXLANs can be added to a single VLAN-aware bridge. A one to one association is used between the VXLAN VNI and the VLAN, with the bridge access VLAN definition on the VXLAN and the VLAN membership definition on the local bridge member interfaces.

The configuration example below shows the differences between a VXLAN configured for traditional bridge mode and one configured for VLAN-aware mode. The configurations use head end replication (HER) together with the VLAN-aware bridge to map VLANs to VNIs.

See VXLAN Scale for information about the number of VXLANs you can configure simultaneously.

...
auto lo
iface lo inet loopback
    address 10.35.0.10/32

auto bridge
iface bridge
    bridge-ports uplink
    bridge-pvid 1
    bridge-vids 1-100
    bridge-vlan-aware yes
auto vni-10000
iface vni-10000
    alias CUSTOMER X VLAN 10
    bridge-access 10
    vxlan-id 10000
    vxlan-local-tunnelip 10.35.0.10
    vxlan-remoteip 10.35.0.34
...

Configure a Static MAC Address Entry

You can add a static MAC address entry to the layer 2 table for an interface within the VLAN-aware bridge by running a command similar to the following:

cumulus@switch:~$ sudo bridge fdb add 12:34:56:12:34:56 dev swp1 vlan 150 master static
cumulus@switch:~$ sudo bridge fdb show
44:38:39:00:00:7c dev swp1 master bridge permanent
12:34:56:12:34:56 dev swp1 vlan 150 master bridge static
44:38:39:00:00:7c dev swp1 self permanent
12:12:12:12:12:12 dev swp1 self permanent
12:34:12:34:12:34 dev swp1 self permanent
12:34:56:12:34:56 dev swp1 self permanent
12:34:12:34:12:34 dev bridge master bridge permanent
44:38:39:00:00:7c dev bridge vlan 500 master bridge permanent
12:12:12:12:12:12 dev bridge master bridge permanent

Caveats and Errata

Spanning Tree Protocol (STP)

IGMP Snooping

IGMP snooping and group membership are supported on a per-VLAN basis; however, the IGMP snooping configuration (including enable, disable, and mrouter ports) is defined on a per-bridge port basis.

VLAN Translation

A bridge in VLAN-aware mode cannot have VLAN translation enabled. Only traditional mode bridges can utilize VLAN translation.

Convert Bridges between Supported Modes

You cannot convert traditional mode bridges automatically to and from a VLAN-aware bridge. You must delete the original configuration and bring down all member switch ports before creating a new bridge.

Traditional Bridge Mode

Cumulus Networks recommends you use a VLAN-aware bridge on your switch. Use traditional mode bridges only if you need to run more than one bridge on the switch or if you need to use PVSTP+.

Configure a Traditional Mode Bridge

The following examples show how to create a simple traditional mode bridge configuration on the switch. The example also shows some optional elements:

To configure spanning tree options for a bridge interface, refer to Spanning Tree and Rapid Spanning Tree.

NCLU Commands

The following example commands configure a traditional mode bridge called my_bridge with IP address 10.10.10.10/24. swp1, swp2, swp3, and swp4 are members of the bridge.

cumulus@switch:~$ net add bridge my_bridge ports swp1-4
cumulus@switch:~$ net add bridge my_bridge ip address 10.10.10.10/24
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. The following example command configures a traditional mode bridge called my_bridge with IP address 10.10.10.10/24. swp1, swp2, swp3, and swp4 are members of the bridge.

...
auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4

auto my_bridge 
iface my_bridge 
    address 10.10.10.10/24 
    bridge-ports swp1 swp2 swp3 swp4 
    bridge-vlan-aware no
...

Run the ifreload -a command to reload the network configuration:

cumulus@switch:~$ sudo ifreload -a

The name of the bridge must be:

  • Compliant with Linux interface naming conventions.
  • Unique within the switch.
  • Something other than bridge, **** as Cumulus Linux reserves that name for a single VLAN-aware bridge.

Do not try to bridge the management port, eth0, with any switch ports (swp0, swp1, and so on). For example, if you create a bridge with eth0 and swp1, it does not work.

Configure Multiple Traditional Mode Bridges

You can configure multiple bridges to logically divide a switch into multiple layer 2 domains. This allows for hosts to communicate with other hosts in the same domain, while separating them from hosts in other domains.

The diagram below shows a multiple bridge configuration, where host-1 and host-2 are connected to bridge-A, while host-3 and host-4 are connected to bridge-B:

This example configuration looks like this in the /etc/network/interfaces file:

...
auto bridge-A
iface bridge-A
    bridge-ports swp1 swp2
    bridge-vlan-aware no

auto bridge-B
iface bridge-B
    bridge-ports swp3 swp4
    bridge-vlan-aware no
...

Trunks in Traditional Bridge Mode

The standard for trunking is 802.1Q. The 802.1Q specification adds a 4 byte header within the Ethernet frame that identifies the VLAN of which the frame is a member.

802.1Q also identifies an untagged frame as belonging to the native VLAN (most network devices default their native VLAN to 1). The concept of native, non-native, tagged or untagged has generated confusion due to mixed terminology and vendor-specific implementations. In Cumulus Linux:

A bridge in traditional mode has no concept of trunks, just tagged or untagged frames. With a trunk of 200 VLANs, there would need to be 199 bridges, each containing a tagged physical interface, and one bridge containing the native untagged VLAN. See the examples below for more information.

The interaction of tagged and un-tagged frames on the same trunk often leads to undesired and unexpected behavior. A switch that uses VLAN 1 for the native VLAN may send frames to a switch that uses VLAN 2 for the native VLAN, thus merging those two VLANs and their spanning tree state.

Trunk Example

To create the above example, add the following configuration to the /etc/network/interfaces file:

...
auto br-VLAN100
iface br-VLAN100
   bridge-ports swp1.100 swp2.100

auto br-VLAN200
iface br-VLAN200
   bridge-ports swp1.200 swp2.200
...

VLAN Tagging Examples

You can find more examples of VLAN tagging in the VLAN tagging chapter.

Configure ARP Timers

Cumulus Linux does not often interact directly with end systems as much as end systems interact with one another. Therefore, after a successful address resolution protocol (ARP) places a neighbor into a reachable state, Cumulus Linux might not interact with the client again for a long enough period of time for the neighbor to move into a stale state. To keep neighbors in the reachable state, Cumulus Linux includes a background process (/usr/bin/neighmgrd). The background process tracks neighbors that move into a stale, delay, or probe state, and attempts to refresh their state before they are removed from the Linux kernel and from hardware forwarding. The neighmgrd process only adds a neighbor if the sender’s IP in the ARP packet is in one of the SVI’s subnets (you can disable this check by setting subnet_checks to 0 in the /etc/cumulus/neighmgr.conf file).

The ARP refresh timer defaults to 1080 seconds (18 minutes). To change this setting, follow the procedures outlined in this knowledge base article.

Caveats

On Broadcom switches, when two VLAN subinterfaces are bridged to each other in a traditional mode bridge, switchd does not assign an internal resource ID to the subinterface, which is expected for each VLAN subinterface. To work around this issue, add a VXLAN on the bridge so that it does not require a real tunnel IP address.

VLAN Tagging

This topic shows two examples of VLAN tagging, one basic and one more advanced. They both demonstrate the streamlined interface configuration from ifupdown2.

VLAN Tagging, a Basic Example

A simple configuration demonstrating VLAN tagging involves two hosts connected to a switch.

To configure the above example, edit the /etc/network/interfaces file and add a configuration like the following:

# Config for host1

auto swp1
iface swp1

auto swp1.100
iface swp1.100

# Config for host2
# swp2 must exist to create the .1Q subinterfaces, but it is not assigned an address

auto swp2
iface swp2

auto swp2.120
iface swp2.120

auto swp2.130
iface swp2.130

VLAN Tagging, an Advanced Example

This example of VLAN tagging is more complex, involving three hosts and two switches, with a number of bridges and a bond connecting them all.

Although not explicitly designated, the bridge member ports function as 802.1Q access ports and trunk ports. In the example above, comparing Cumulus Linux with a traditional Cisco device:

To create the above configuration, edit the /etc/network/interfaces file and add a configuration like the following:

# Config for host1

# swp1 does not need an iface section unless it has a specific setting,
# it will be picked up as a dependent of swp1.100.
# And swp1 must exist in the system to create the .1q subinterfaces..
# but it is not applied to any bridge..or assigned an address.

auto swp1.100
iface swp1.100

# Config for host2
# swp2 does not need an iface section unless it has a specific setting,
# it will be picked up as a dependent of swp2.100 and swp2.120.
# And swp2 must exist in the system to create the .1q subinterfaces..
# but it is not applied to any bridge..or assigned an address.

auto swp2.100
iface swp2.100

auto swp2.120
iface swp2.120

# Config for host3
# swp3 does not need an iface section unless it has a specific setting,
# it will be picked up as a dependent of swp3.120 and swp3.130.
# And swp3 must exist in the system to create the .1q subinterfaces..
# but it is not applied to any bridge..or assigned an address.

auto swp3.120
iface swp3.120

auto swp3.130
iface swp3.130

# Configure the bond

auto bond2
iface bond2
  bond-slaves glob swp4-7

# configure the bridges

auto br-untagged
iface br-untagged
    address 10.0.0.1/24
    bridge-ports swp1 bond2
    bridge-stp on

auto br-tag100
iface br-tag100
    address 10.0.100.1/24
    bridge-ports swp1.100 swp2.100 bond2.100
    bridge-stp on

auto br-vlan120
iface br-vlan120
    address 10.0.120.1/24
    bridge-ports swp2.120 swp3.120 bond2.120
    bridge-stp on

auto v130
iface v130
    address 10.0.130.1/24
    bridge-ports swp3.130 bond2.130
    bridge-stp on

#

To verify:

cumulus@switch:~$ sudo mstpctl showbridge br-tag100
br-tag100 CIST info
  enabled         yes
  bridge id       8.000.44:38:39:00:32:8B
  designated root 8.000.44:38:39:00:32:8B
  regional root   8.000.44:38:39:00:32:8B
  root port       none
  path cost     0          internal path cost   0
  max age       20         bridge max age       20
  forward delay 15         bridge forward delay 15
  tx hold count 6          max hops             20
  hello time    2          ageing time          300
  force protocol version     rstp
  time since topology change 333040s
  topology change count      1
  topology change            no
  topology change port       swp2.100
  last topology change port  None
cumulus@switch:~$ sudo mstpctl showportdetail br-tag100  | grep -B 2 state
br-tag100:bond2.100 CIST info
  enabled            yes                     role                 Designated
  port id            8.003                   state                forwarding
--
br-tag100:swp1.100 CIST info
  enabled            yes                     role                 Designated
  port id            8.001                   state                forwarding
--
  br-tag100:swp2.100 CIST info
  enabled            yes                     role                 Designated
  port id            8.002                   state                forwarding
cumulus@switch:~$ cat /proc/net/vlan/config
VLAN Dev name    | VLAN ID
Name-Type: VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD
bond2.100      | 100  | bond2
bond2.120      | 120  | bond2
bond2.130      | 130  | bond2
swp1.100       | 100  | swp1
swp2.100       | 100  | swp2
swp2.120       | 120  | swp2
swp3.120       | 120  | swp3
swp3.130       | 130  | swp3
cumulus@switch:~$ cat /proc/net/bonding/bond2
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)

Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer3+4 (1)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

802.3ad info
LACP rate: fast
Min links: 0
Aggregator selection policy (ad_select): stable
Active Aggregator Info:
    Aggregator ID: 3
    Number of ports: 4
    Actor Key: 33
    Partner Key: 33
    Partner Mac Address: 44:38:39:00:32:cf

Slave Interface: swp4
MII Status: up
Speed: 10000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:32:8e
Aggregator ID: 3
Slave queue ID: 0

Slave Interface: swp5
MII Status: up
Speed: 10000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:32:8f
Aggregator ID: 3
Slave queue ID: 0

Slave Interface: swp6
MII Status: up
Speed: 10000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:32:90
Aggregator ID: 3
Slave queue ID: 0

Slave Interface: swp7
MII Status: up
Speed: 10000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 44:38:39:00:32:91
Aggregator ID: 3
Slave queue ID: 0

A single bridge cannot contain multiple subinterfaces of the same port as members. Attempting to apply such a configuration will result in an error:

cumulus@switch:~$ sudo brctl addbr another_bridge
cumulus@switch:~$ sudo brctl addif another_bridge swp9 swp9.100
bridge cannot contain multiple subinterfaces of the same port: swp9, swp9.100

VLAN Translation

By default, Cumulus Linux does not allow VLAN subinterfaces associated with different VLAN IDs to be part of the same bridge. Base interfaces are not explicitly associated with any VLAN IDs and are exempt from this restriction.

In some cases, it may be useful to relax this restriction. For example, two servers might be connected to the switch using VLAN trunks, but the VLAN numbering provisioned on the two servers are not consistent. You can choose to just bridge two VLAN subinterfaces of different VLAN IDs from the servers. You do this by enabling the sysctl net.bridge.bridge-allow-multiple-vlans. Packets entering a bridge from a member VLAN subinterface will egress another member VLAN subinterface with the VLAN ID translated.

A bridge in VLAN-aware mode cannot have VLAN translation enabled for it; only bridges configured in traditional mode can utilize VLAN translation.

The following example enables the VLAN translation sysctl:

cumulus@switch:~$ echo net.bridge.bridge-allow-multiple-vlans = 1 | sudo tee /etc/sysctl.d/multiple_vlans.conf
net.bridge.bridge-allow-multiple-vlans = 1
cumulus@switch:~$ sudo sysctl -p /etc/sysctl.d/multiple_vlans.conf
net.bridge.bridge-allow-multiple-vlans = 1

If the sysctl is enabled and you want to disable it, run the above example, setting the sysctl net.bridge.bridge-allow-multiple-vlans to 0.

After sysctl is enabled, ports with different VLAN IDs can be added to the same bridge. In the following example, packets entering the bridge br-mix from swp10.100 will be bridged to swp11.200 with the VLAN ID translated from 100 to 200:

cumulus@switch:~$ sudo brctl addif br_mix swp10.100 swp11.200

cumulus@switch:~$ sudo brctl show br_mix
bridge name     bridge id               STP enabled     interfaces
br_mix          8000.4438390032bd       yes             swp10.100
                                                        swp11.200

Multi-Chassis Link Aggregation - MLAG

Multi-Chassis Link Aggregation (MLAG) enables a server or switch with a two-port bond, such as a link aggregation group/LAG, EtherChannel, port group or trunk, to connect those ports to different switches and operate as if they are connected to a single, logical switch. This provides greater redundancy and greater system throughput.

MLAG or CLAG: The Cumulus Linux implementation of MLAG is referred to by other vendors as CLAG, MC-LAG or VPC. You will even see references to CLAG in CumulusLinux, including the management daemon, named clagd, and other options in the code, such as clag-id, which exist for historical purposes. The Cumulus Linux implementation is truly a multi-chassis link aggregation protocol, so we call it MLAG.

Dual-connected devices can create LACP bonds that contain links to each physical switch. Therefore, active-active links from the dual-connected devices are supported even though they are connected to two different physical switches.

A basic setup looks like this:

You can see an example of how to set up this configuration by running:

cumulus@switch:~$ net example clag basic-clag

The two switches, S1 and S2, known as peer switches, appear as a single device to the bond on host H1. H1 distributes traffic between the two links to S1 and S2 in any way that you configure on the host. Similarly, traffic inbound to H1 can traverse S1 or S2 and arrive at H1.

MLAG Requirements

MLAG has these requirements:

More elaborate configurations are also possible. The number of links between the host and the switches can be greater than two and does not have to be symmetrical:

Additionally, because S1 and S2 appear as a single switch to other bonding devices, you can also connect pairs of MLAG switches to each other in a switch-to-switch MLAG configuration:

In the above example, L1 and L2 are also MLAG peer switches and present a two-port bond from a single logical system to S1 and S2. S1 and S2 do the same as far as L1 and L2 are concerned. For a switch-to-switch MLAG configuration, each switch pair must have a unique system MAC address. In the example, switch L1 and L2 each have the same system MAC address. Switch pair S1 and S2 each have the same system MAC address; however, it is a different system MAC address than the one used by the switch pair L1 and L2.

LACP and Dual-Connectedness

For MLAG to operate correctly, the peer switches must know which links are dual-connected or are connected to the same host or switch. You must specify a clag-id for every dual-connected bond on each peer switch; the clag-id must be the same for the corresponding bonds on both peer switches. Typically, Link Aggregation Control Protocol (LACP), the IEEE standard protocol for managing bonds, is used for verifying dual-connectedness. LACP runs on the dual-connected device and on each of the peer switches. On the dual-connected device, the only configuration requirement is to create a bond that is managed by LACP.

However, if you cannot use LACP in your environment, you can configure the bonds in balance-xor mode. When using balance-xor mode to dual-connect host-facing bonds in an MLAG environment, you must configure the clag-id parameter on the MLAG bonds, which must be the same on both MLAG switches. Otherwise, the bonds are treated by the MLAG switch pair as if they are single-connected. Dual-connectedness is solely determined by matching clag-id and any misconnection is not detected.

On each of the peer switches, you must place the links that are connected to the dual-connected host or switch in the bond. This is true even if the links are a single port on each peer switch, where each port is placed into a bond, as shown below:

All of the dual-connected bonds on the peer switches have their system ID set to the MLAG system ID. Therefore, from the point of view of the hosts, each of the links in its bond is connected to the same system and so the host uses both links.

Each peer switch periodically makes a list of the LACP partner MAC addresses for all of their bonds and sends that list to its peer (using the clagd service; see below). The LACP partner MAC address is the MAC address of the system at the other end of a bond (hosts H1, H2, and H3 in the figure above). When a switch receives this list from its peer, it compares the list to the LACP partner MAC addresses on its switch. If any matches are found and the clag-id for those bonds match, then that bond is a dual-connected bond. You can find the LACP partner MAC address by the running net show bridge macs command or by examining the /sys/class/net/<bondname>/bonding/ad_partner_mac sysfs file for each bond.

Configure MLAG

To configure MLAG, you need to:

MLAG synchronizes the dynamic state between the two peer switches but it does not synchronize the switch configurations. After modifying the configuration of one peer switch, you must make the same changes to the configuration on the other peer switch. This applies to all configuration changes, including:

To verify VLAN membership configuration, run the NCLU net show clag verify-vlans verbose command or the Linux clagctl -v verifyvlans command. For example:

cumulus@switch:~$ net show clag verify-vlans verbose
Our Bond Interface   VlanId   Peer Bond Interface
------------------   ------   -------------------
server01                  1   server01
server01                 10   server01
server01                 20   server01
server01                 30   server01
server01                 40   server01
server01                 50   server01
uplink                    1   uplink
uplink                   10   uplink
uplink                   20   uplink
uplink                   30   uplink
uplink                   40   uplink
uplink                   50   uplink
uplink                  100   uplink
uplink                  101   uplink
uplink                  102   uplink
uplink                  103   uplink
uplink                  104   uplink
uplink                  105   uplink
...

Reserved MAC Address Range

To prevent MAC address conflicts with other interfaces in the same bridged network, Cumulus Networks has reserved a range of MAC addresses specifically to use with MLAG. This range of MAC addresses is 44:38:39:ff:00:00 to 44:38:39:ff:ff:ff.

Cumulus Networks recommends you use this range of MAC addresses when configuring MLAG.

  • You cannot use the same MAC address for different MLAG pairs. Make sure you specify a different clagd-sys-mac setting for each MLAG pair in the network.
  • If you configure MLAG with NCLU commands, Cumulus Linux does not check against a possible collision with VLANs outside the default reserved range when creating the peer link interfaces, in case the reserved VLAN range has been modified.

Configure the Host or Switch

On your dual-connected device, create a bond that uses LACP. The method you use varies with the type of device you are configuring. The following image is a basic MLAG configuration, showing all the essential elements; a more detailed two-leaf/two-spine configuration is shown below.

Configure the Interfaces

Place every interface that connects to the MLAG pair from a dual-connected device into a bond, even if the bond contains only a single link on a single physical switch (even though the MLAG pair contains two or more links). Layer 2 data travels over this bond. In the examples throughout this chapter, peerlink is the name of the bond.

Single-attached hosts, also known as orphan ports, can be just a member of the bridge.

Additionally, configure the fast mode of LACP on the bond to allow more timely updates of the LACP state. These bonds are then placed in a bridge, which must include the peer link between the switches.

To enable communication between the clagd services on the peer switches, do the following:

For example, if peerlink is the inter-chassis bond, and VLAN 4094 is the peer link VLAN, configure peerlink.4094 as follows:

NCLU Commands
cumulus@switch:~$ net add clag peer sys-mac 44:38:39:FF:40:94 interface swp49-50 linklocal backup-ip 192.0.2.50
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Do not add VLAN 4094 to the bridge VLAN list; VLAN 4094 for the peer link subinterface cannot also be configured as a bridged VLAN with bridge VIDs under the bridge.

To enable MLAG, you must add peerlink to a traditional or VLAN-aware bridge. The commands below add peerlink to a VLAN-aware bridge:

cumulus@switch:~$ net add bridge bridge ports peerlink
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file to add the peer link.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.0.2.50
    clagd-peer-ip linklocal
    clagd-sys-mac 44:38:39:FF:40:94
...

To enable MLAG, you must add peerlink to a traditional or VLAN-aware bridge. The configuration below adds peerlink to a VLAN-aware bridge:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports peerlink
    bridge-vlan-aware yes
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

When you change the MLAG configuration in the interfaces file, the changes take effect when you bring the peer link interface up with ifup or ifreload -a. Do not use systemctl restart clagd.service to apply the new configuration.

Do not use 169.254.0.1 as the MLAG peer link IP address; Cumulus Linux uses this address exclusively for BGP unnumbered interfaces.

Switch Roles and Priority Setting

Each MLAG-enabled switch in the pair has a role. When the peering relationship is established between the two switches, one switch is put into the primary role, and the other into the secondary role. When an MLAG-enabled switch is in the secondary role, it does not send STP BPDUs on dual-connected links; it only sends BPDUs on single-connected links. The switch in the primary role sends STP BPDUs on all single- and dual-connected links.

Sends BPDUs ViaPrimarySecondary
Single-connected linksYesYes
Dual-connected linksYesNo

By default, the role is determined by comparing the MAC addresses of the two sides of the peering link; the switch with the lower MAC address assumes the primary role. You can override this by setting the clagd-priority option for the peer link:

NCLU Commands

The following command example sets the clagd-priority option for the peer link.

cumulus@switch:~$ net add interface peerlink.4094 clag priority 2048
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add the clagd-priority option. The following example sets the clagd-priority option for the peer link:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink.4094
iface peerlink.4094
    clagd-peer-ip linklocal
    clagd-backup-ip 192.0.2.50
    clagd-sys-mac 44:38:39:FF:40:94
    clagd-priority 2048
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

The switch with the lower priority value is given the primary role; the default value is 32768 and the range is 0 to 65535. Read the clagd(8) and clagctl(8) man pages for more information.

When the clagd service exits during switch reboot or if you stop the service on the primary switch, the peer switch that is in the secondary role becomes the primary.

However, if the primary switch goes down without stopping the clagd service for any reason, or if the peer link goes down, the secondary switch does not change its role. In case the peer switch is determined to be not alive, the switch in the secondary role rolls back the LACP system ID to be the bond interface MAC address instead of the clagd-sys-mac and the switch in primary role uses the clagd-sys-mac as the LACP system ID on the bonds.

clagctl Timers

The clagd service has a number of timers that you can tune for enhanced performance. The relevant timers are:

To set a timer, use NCLU. For example, to set the peerTimeout to 900 seconds:

cumulus@switch:~$ net add interface peerlink.4094 clag args --peerTimeout 900
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

You can run clagctl params to see the settings for all of the clagd parameters.

cumulus@leaf01:~$ clagctl params
clagVersion = 1.3.0
clagDataVersion = 1.3.0
clagCmdVersion = 1.1.0
peerIp = 169.254.1.2
peerIf = peerlink.4094
sysMac = 44:38:39:ff:00:01
lacpPoll = 2
currLacpPoll = 2
peerConnect = 1
cmdConnect = 1
peerLinkPoll = 1
switchdReadyTimeout = 120
reloadTimer = 300
periodicRun = 4
priority = 1000
quiet = False
debug = 0x0
verbose = False
log = syslog
vm = True
peerPort = 5342
peerTimeout = 20
initDelay = 180
sendTimeout = 30
sendBufSize = 65536
forceDynamic = False
dormantDisable = False
redirectEnable = False
backupIp = 192.168.0.12
backupVrf = None
backupPort = 5342
vxlanAnycast = None
neighSync = True
permanentMacSync = True
cmdLine = /usr/sbin/clagd --daemon 169.254.1.2 peerlink.4094 44:38:39:FF:00:01 --priority 1000 --backupIp 192.168.0.12 --peerTimeout 900
peerlinkLearnEnable = False
cumulus@leaf01:~$

Example MLAG Configuration

The example configuration below configures two bonds for MLAG, each with a single port, a peer link that is a bond with two member ports, and three VLANs on each port.

You can see a more traditional layer 2 example configuration in NCLU; run net example clag l2-with-server-vlan-trunks. For a very basic configuration with just one pair of switches and a single host, run net example clag l2-with-server-vlan-trunks.

You configure these interfaces using NCLU, so the bridges are in VLAN-aware mode. The bridges use these Cumulus Linux-specific keywords:

The bridge configurations below indicate that each bond carries tagged frames on VLANs 10, 20, 30, 40, 50, and 100 to 200 (as specified by bridge-vids), but untagged frames on VLAN 1 (as specified by bridge-pvid). Also, take note on how you configure the VLAN subinterfaces used for clagd communication (peerlink.4094 in the sample configuration below). Finally, the host configurations for server01 through server04 are not shown here. The configurations for each corresponding node are almost identical, except for the IP addresses used for managing the clagd service.

Make sure that the VLAN subinterface is not in your layer 2 domain and does not have a very high VLAN ID (up to 4094). Read more about the range of VLAN IDs you can use.

The commands to create the configurations for both spines look like the following. Note that the clag-id and clagd-sys-mac must be the same for the corresponding bonds on spine01 and spine02:

spine01
cumulus@spine01:~$ net show configuration commands
net add interface swp1-4
net add loopback lo ip address 10.0.0.21/32
net add interface eth0 ip address dhcp

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@spine01:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.21/32

auto eth0
iface eth0 inet dhcp

# downlinks
auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4
spine02
cumulus@spine02:~$ net show configuration commands
net add interface swp1-4
net add loopback lo ip address 10.0.0.22/32
net add interface eth0 ip address dhcp

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@spine02:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.22/32

auto eth0
iface eth0 inet dhcp

# downlinks
auto swp1
iface swp1

auto swp2
iface swp2

auto swp3
iface swp3

auto swp4
iface swp4

Here is an example configuration for the switches leaf01 through leaf04. Note that the clag-id and clagd-sys-mac must be the same for the corresponding bonds on leaf01 and leaf02 as well as leaf03 and leaf04:

leaf01
cumulus@leaf01:~$ net show configuration commands
net add loopback lo ip address 10.0.0.11/32
net add bgp autonomous-system 65011
net add bgp router-id 10.0.0.11
net add bgp ipv4 unicast network 10.0.0.11/32
net add routing prefix-list ipv4 dc-leaf-in seq 10 permit 0.0.0.0/0
net add routing prefix-list ipv4 dc-leaf-in seq 20 permit 10.0.0.0/24 le 32
net add routing prefix-list ipv4 dc-leaf-in seq 30 permit 172.16.2.0/24
net add routing prefix-list ipv4 dc-leaf-out seq 10 permit 172.16.1.0/24
net add bgp neighbor fabric peer-group
net add bgp neighbor fabric remote-as external
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-in in
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-out out
net add bgp neighbor swp51-52 interface peer-group fabric
net add vlan 100 ip address 172.16.1.1/24
net add bgp ipv4 unicast network 172.16.1.1/24
net add clag peer sys-mac 44:38:39:FF:00:01 interface swp49-50 primary backup-ip 192.168.1.12
net add clag port bond server1 interface swp1 clag-id 1
net add clag port bond server2 interface swp2 clag-id 2
net add bond server1-2 bridge access 100
net add bond server1-2 stp portadminedge
net add bond server1-2 stp bpduguard

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@leaf01:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.11/32

auto eth0
iface eth0 inet dhcp

auto swp1
iface swp1

auto swp2
iface swp2

# peerlink
auto swp49
iface swp49
    post-up ip link set $IFACE promisc on     # Only required on VX

auto swp50
iface swp50
    post-up ip link set $IFACE promisc on     # Only required on VX

# uplinks
auto swp51
iface swp51

auto swp52
iface swp52

# bridge to hosts
auto bridge
iface bridge
    bridge-ports peerlink server1 server2
    bridge-vids 100
    bridge-vlan-aware yes

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.168.1.12
    clagd-peer-ip linklocal
    clagd-priority 1000
    clagd-sys-mac 44:38:39:FF:00:01

auto server1
iface server1
    bond-slaves swp1
    bridge-access 100
    clag-id 1
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto server2
iface server2
    bond-slaves swp2
    bridge-access 100
    clag-id 2
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto vlan100
iface vlan100
    address 172.16.1.1/24
    vlan-id 100
    vlan-raw-device bridge
leaf02
cumulus@leaf02:~$ net show conf commands
net add loopback lo ip address 10.0.0.12/32
net add bgp autonomous-system 65012
net add bgp router-id 10.0.0.12
net add bgp ipv4 unicast network 10.0.0.12/32
net add routing prefix-list ipv4 dc-leaf-in seq 10 permit 0.0.0.0/0
net add routing prefix-list ipv4 dc-leaf-in seq 20 permit 10.0.0.0/24 le 32
net add routing prefix-list ipv4 dc-leaf-in seq 30 permit 172.16.2.0/24
net add routing prefix-list ipv4 dc-leaf-out seq 10 permit 172.16.1.0/24
net add bgp neighbor fabric peer-group
net add bgp neighbor fabric remote-as external
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-in in
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-out out
net add bgp neighbor swp51-52 interface peer-group fabric
net add vlan 100 ip address 172.16.1.2/24
net add bgp ipv4 unicast network 172.16.1.2/24
net add clag peer sys-mac 44:38:39:FF:00:01 interface swp49-50 secondary backup-ip 192.168.1.11
net add clag port bond server1 interface swp1 clag-id 1
net add clag port bond server2 interface swp2 clag-id 2
net add bond server1-2 bridge access 100
net add bond server1-2 stp portadminedge
net add bond server1-2 stp bpduguard

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@leaf02:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.12/32

auto eth0
iface eth0 inet dhcp

auto swp1
iface swp1

auto swp2
iface swp2

# peerlink
auto swp49
iface swp49
    post-up ip link set $IFACE promisc on     # Only required on VX

auto swp50
iface swp50
    post-up ip link set $IFACE promisc on     # Only required on VX

# uplinks
auto swp51
iface swp51

auto swp52
iface swp52

# bridge to hosts
auto bridge
iface bridge
    bridge-ports peerlink server1 server2
    bridge-vids 100
    bridge-vlan-aware yes

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.168.1.11
    clagd-peer-ip linklocal
    clagd-sys-mac 44:38:39:FF:00:01

auto server1
iface server1
    bond-slaves swp1
    bridge-access 100
    clag-id 1
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto server2
iface server2
    bond-slaves swp2
    bridge-access 100
    clag-id 2
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes
 
auto vlan100
iface vlan100
    address 172.16.1.2/24
    vlan-id 100
    vlan-raw-device bridge
leaf03
cumulus@leaf03:~$ net show conf commands
net add loopback lo ip address 10.0.0.13/32
net add bgp autonomous-system 65013
net add bgp router-id 10.0.0.13
net add bgp ipv4 unicast network 10.0.0.13/32
net add routing prefix-list ipv4 dc-leaf-in seq 10 permit 0.0.0.0/0
net add routing prefix-list ipv4 dc-leaf-in seq 20 permit 10.0.0.0/24 le 32
net add routing prefix-list ipv4 dc-leaf-in seq 30 permit 172.16.2.0/24
net add routing prefix-list ipv4 dc-leaf-out seq 10 permit 172.16.1.0/24
net add bgp neighbor fabric peer-group
net add bgp neighbor fabric remote-as external
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-in in
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-out out
net add bgp neighbor swp51-52 interface peer-group fabric
net add vlan 100 ip address 172.16.1.3/24
net add bgp ipv4 unicast network 172.16.1.3/24
net add clag peer sys-mac 44:38:39:FF:00:02 interface swp49-50 primary backup-ip 192.168.1.14
net add clag port bond server3 interface swp1 clag-id 3
net add clag port bond server4 interface swp2 clag-id 4
net add bond server3-4 bridge access 100
net add bond server3-4 stp portadminedge
net add bond server3-4 stp bpduguard

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@leaf03:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.13/32

auto eth0
iface eth0 inet dhcp

auto swp1
iface swp1

auto swp2
iface swp2

# peerlink
auto swp49
iface swp49
    post-up ip link set $IFACE promisc on     # Only required on VX

auto swp50
iface swp50
    post-up ip link set $IFACE promisc on     # Only required on VX

# uplinks
auto swp51
iface swp51

auto swp52
iface swp52

# bridge to hosts
auto bridge
iface bridge
    bridge-ports peerlink server3 server4
    bridge-vids 100
    bridge-vlan-aware yes

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.168.1.14
    clagd-peer-ip linklocal
    clagd-priority 1000
    clagd-sys-mac 44:38:39:FF:00:02

auto server3
iface server3
    bond-slaves swp1
    bridge-access 100
    clag-id 3
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto server4
iface server4
    bond-slaves swp2
    bridge-access 100
    clag-id 4
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto vlan100
iface vlan100
    address 172.16.1.3/24
    vlan-id 100
    vlan-raw-device bridge
leaf04
cumulus@leaf04:~$ net show configuration commands
net add loopback lo ip address 10.0.0.14/32
net add bgp autonomous-system 65014
net add bgp router-id 10.0.0.14
net add bgp ipv4 unicast network 10.0.0.14/32
net add routing prefix-list ipv4 dc-leaf-in seq 10 permit 0.0.0.0/0
net add routing prefix-list ipv4 dc-leaf-in seq 20 permit 10.0.0.0/24 le 32
net add routing prefix-list ipv4 dc-leaf-in seq 30 permit 172.16.2.0/24
net add routing prefix-list ipv4 dc-leaf-out seq 10 permit 172.16.1.0/24
net add bgp neighbor fabric peer-group
net add bgp neighbor fabric remote-as external
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-in in
net add bgp ipv4 unicast neighbor fabric prefix-list dc-leaf-out out
net add bgp neighbor swp51-52 interface peer-group fabric
net add vlan 100 ip address 172.16.1.4/24
net add bgp ipv4 unicast network 172.16.1.4/24
net add clag peer sys-mac 44:38:39:FF:00:02 interface swp49-50 secondary backup-ip 192.168.1.13
net add clag port bond server3 interface swp1 clag-id 3
net add clag port bond server4 interface swp2 clag-id 4
net add bond server3-4 bridge access 100
net add bond server3-4 stp portadminedge
net add bond server3-4 stp bpduguard

These commands create the following configuration in the /etc/network/interfaces file:

cumulus@leaf04:~$ cat /etc/network/interfaces
auto lo
iface lo inet loopback
    address 10.0.0.14/32

auto eth0
iface eth0 inet dhcp

auto swp1
iface swp1

auto swp2
iface swp2

# peerlink
auto swp49
iface swp49
    post-up ip link set $IFACE promisc on     # Only required on VX

auto swp50
iface swp50
    post-up ip link set $IFACE promisc on     # Only required on VX

# uplinks
auto swp51
iface swp51

auto swp52
iface swp52

# bridge to hosts
auto bridge
iface bridge
    bridge-ports peerlink server3 server4
    bridge-vids 100
    bridge-vlan-aware yes

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.168.1.13
    clagd-peer-ip linklocal
    clagd-sys-mac 44:38:39:FF:00:02

auto server3
iface server3
    bond-slaves swp1
    bridge-access 100
    clag-id 3
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto server4
iface server4
    bond-slaves swp2
    bridge-access 100
    clag-id 4
    mstpctl-bpduguard yes
    mstpctl-portadminedge yes

auto vlan100
iface vlan100
    address 172.16.1.4/24
    vlan-id 100
    vlan-raw-device bridge

Disable clagd on an Interface

In the configurations above, the clagd-peer-ip and clagd-sys-mac parameters are mandatory, while the rest are optional. When mandatory clagd commands are present under a peer link subinterface, the clagd-enable option is not present but is enabled by default. To disable clagd on the subinterface, set clagd-enable to no:

NCLU Commands
cumulus@switch:~$ net add interface peerlink.4094 clag enable no
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add clagd-enable no to the interface stanza:

cumulus@switch:~$ sudo nano /etc/network/interfaces 
...
auto peerlink.4094
iface peerlink.4094
    clagd-backup-ip 192.168.1.12
    clagd-enable no
    clagd-peer-ip linklocal
    clagd-priority 1000
    clagd-sys-mac 44:38:39:FF:00:01
...

Run the ifreload -a command to reload the configuration.

cumulus@switch:~$ sudo ifreload -a

Use clagd-priority to set the role of the MLAG peer switch to primary or secondary. Each peer switch in an MLAG pair must have the same clagd-sys-mac setting. Each clagd-sys-mac setting must be unique to each MLAG pair in the network. For more details, refer to man clagd.

Check the MLAG Configuration Status

To check the status of your MLAG configuration, run the NCLU net show clag command or the Linux clagctl command. For example:

cumulus@switch:~$ net show clag 
The peer is alive
    Peer Priority, ID, and Role: 4096 44:38:39:FF:00:01 primary
     Our Priority, ID, and Role: 8192 44:38:39:FF:00:02 secondary
          Peer Interface and IP: peerlink.4094 linklocal
                      Backup IP: 192.168.1.12 (inactive)
                     System MAC: 44:38:39:FF:00:01

CLAG Interfaces
Our Interface      Peer Interface     CLAG Id   Conflicts              Proto-Down Reason
----------------   ----------------   -------   --------------------   -----------------
         server1   server1            1         -                      -
         server2   server2            2         -                      -

Configure MLAG with a Bridge in Traditional Mode

To configure MLAG with a traditional mode bridge instead of VLAN-aware mode, the peer link and all dual-connected links must be configured as untagged/native ports on a bridge (note the absence of any VLANs in the bridge-ports line and the lack of the bridge-vlan-aware parameter below):

...
auto br0
iface br0
    bridge-ports peerlink spine1-2 host1 host2
...

The following example shows you how to allow VLAN 100 across the peer link:

...
auto br0.100
iface br0.100
    bridge-ports peerlink.100 bond1.100
    bridge-stp on
...

For a deeper comparison of traditional versus VLAN-aware bridge modes, read this knowledge base article.

In addition to the standard UP and DOWN administrative states, an interface that is a member of an MLAG bond can also be in a protodown state. When MLAG detects a problem that might result in connectivity issues, it can put that interface into protodown state. Such connectivity issues include:

When an interface goes into a protodown state, it results in a local OPER DOWN (carrier down) on the interface.

To show an interface in protodown state, run the NCLU net show bridge link command or the Linux ip link show command. For example:

cumulus@switch:~$ net show bridge link
3: swp1 state DOWN: <NO-CARRIER,BROADCAST,MULTICAST,MASTER,UP> mtu 9216 master pfifo_fast master host-bond1 state DOWN mode DEFAULT qlen 500 protodown on
    link/ether 44:38:39:00:69:84 brd ff:ff:ff:ff:ff:ff

You should specify a backup link for your peer links in case the peer link goes down. When this happens, the clagd service uses the backup link to check the health of the peer switch. The backup link is specified in the clagd-backup-ip parameter.

In an anycast VTEP environment, if you do not specify the clagd-backup-ip parameter, large convergence times (around 5 minutes) can result when the primary MLAG switch is powered off. Then the secondary switch must wait until the reload delay timer expires (which defaults to 300 seconds, or 5 minutes) before bringing up a VNI with its unique loopback IP.

The backup IP address must be different than the peer link IP address (clagd-peer-ip). It must be reachable by a route that does not use the peer link and it must be in the same network namespace as the peer link IP address.

Cumulus Networks recommends you use the switch’s loopback or management IP address for this purpose. Which one should you choose?

To configure a backup link:

NCLU Commands
cumulus@switch:~$ net add interface peerlink.4094 clag backup-ip 192.0.2.50
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

You can also specify the backup UDP port. The port defaults to 5342, but you can configure it with the clagd args --backupPort <PORT> option. For example:

cumulus@switch:~$ net add interface peerlink.4094 clag args --backupPort 5400
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add clag-backup-ip <ip-address> to the peer link configuration. For example:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink.4094
iface peerlink.4094
    netmask 255.255.255.0
    clagd-priority 8192
    clagd-peer-ip linklocal
    clagd-backup-ip 192.0.2.50
    clagd-sys-mac 44:38:39:ff:00:01
    clagd-args --priority 1000
...

You can also specify the backup UDP port. The port defaults to 5342, but you can change the port with clagd-args --backupPort <port> . For example:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink.4094
iface peerlink.4094
    netmask 255.255.255.0
    clagd-priority 8192
    clagd-peer-ip linklocal
    clagd-backup-ip 192.0.2.50
    clagd-args --backupPort 5400
    clagd-sys-mac 44:38:39:ff:00:01
    clagd-args --priority 1000
...

Run ifreload -a to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

To show the backup IP address, run the NCLU net show clag command or the Linux clagctl command. For example:

cumulus@switch:~$ net show clag 
The peer is alive
        Our Priority, ID, and Role: 32768 44:38:39:00:00:41 primary
    Peer Priority, ID, and Role: 32768 44:38:39:00:00:42 secondary
            Peer Interface and IP: peerlink.4094 linklocal
                        Backup IP: 192.168.0.22 (active)
                        System MAC: 44:38:39:FF:40:90

CLAG Interfaces
Our Interface      Peer Interface     CLAG Id   Conflicts              Proto-Down Reason
----------------   ----------------   -------   --------------------   -----------------
        leaf03-04   leaf03-04          1034      -                      -
        exit01-02   -                  2930      -                      -
        leaf01-02   leaf01-02          1012      -                      -

You can configure the backup link to a VRF or management VRF. Include the name of the VRF or management VRF with the clagd-backup-ip command.

You cannot use the VRF on a peer link subinterface.

NCLU Commands
cumulus@switch:~$ net add interface peerlink.4094 clag backup-ip 192.168.0.22 vrf green
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file to include the name of the VRF or management VRF with the clag-backup-ip option. The following configuration links to the management VRF.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto eth0
iface eth0 inet dhcp
        vrf mgmt

auto mgmt
iface mgmt
        vrf-table auto

auto peer-bond.4000
iface peer-bond.4000
        clagd-priority 8192
        clagd-peer-ip linklocal
        clagd-backup-ip 192.0.2.174 vrf mgmt
        clagd-sys-mac 44:38:39:ff:00:01
...

Run ifreload -a to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

To verify the backup link, run the NCLU net show clag command or the Linux clagctl command. For example:

cumulus@switch:~$ net show clag 
The peer is alive
        Our Priority, ID, and Role: 32768 44:38:39:00:00:41 primary
    Peer Priority, ID, and Role: 32768 44:38:39:00:00:42 secondary
            Peer Interface and IP: peerlink.4094 linklocal
                        Backup IP: 192.168.0.22 vrf green (active)
                        System MAC: 44:38:39:FF:40:90

CLAG Interfaces
Our Interface      Peer Interface     CLAG Id   Conflicts              Proto-Down Reason
----------------   ----------------   -------   --------------------   -----------------
        leaf03-04   leaf03-04          1034      -                      -
        exit01-02   -                  2930      -                      -
        leaf01-02   leaf01-02          1012      -                      -

Monitor Dual-Connected Peers

When the switch receives a valid message from its peer, it knows that clagd is alive and executing on that peer. This causes clagd to change the system ID of each bond that is assigned a clag-id from the default value (the MAC address of the bond) to the system ID assigned to both peer switches. This makes the hosts connected to each switch act as if they are connected to the same system so that they use all ports within their bond. Additionally, clagd determines which bonds are dual-connected and modifies the forwarding and learning behavior to accommodate these dual-connected bonds.

If the peer does not receive any messages for three update intervals, that peer switch is assumed to no longer be acting as an MLAG peer. In this case, the switch reverts all configuration changes so that it operates as a standard non-MLAG switch. This includes removing all statically assigned MAC addresses, clearing the egress forwarding mask, and allowing addresses to move from any port to the peer port. After a message is again received from the peer, MLAG operation starts again as described earlier. You can configure a custom timeout setting by adding --peerTimeout <value> to clagd-args:

NCLU Commands

The following example commands set the timeout to 900:

cumulus@switch:~$ net add interface peerlink.4094 clag args --peerTimeout 900
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and add the timeout to the peerlink stanza. The following example sets the timeout to 900:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink.4094
iface peerlink.4094
    clagd-args --backupPort 5400
    clagd-args --peerTimeout 900
    clagd-peer-ip linklocal
    clagd-backup-ip 192.0.2.50
    clagd-priority 8192
    clagd-sys-mac 44:38:39:ff:00:01
...

Run ifreload -a to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

After bonds are identified as dual-connected, clagd sends more information to the peer switch for those bonds. The MAC addresses (and VLANs) that are dynamically learned on those ports are sent along with the LACP partner MAC address for each bond. When a switch receives MAC address information from its peer, it adds MAC address entries on the corresponding ports. As the switch learns and ages out MAC addresses, it informs the peer switch of these changes to its MAC address table so that the peer can keep its table synchronized. Periodically, at 45% of the bridge ageing time, a switch sends its entire MAC address table to the peer, so that peer switch can verify that its MAC address table is properly synchronized.

The switch sends an update frequency value in the messages to its peer, which tells clagd how often the peer will send these messages. You can configure a different frequency by adding --lacpPoll <seconds> to clagd-args:

NCLU Commands

The following example command sets the frequency to 900 seconds:

cumulus@switch:~$ net add interface peerlink.4094 clag args --lacpPoll 900
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. The following example sets the frequency to 900 seconds:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto peerlink.4094
 iface peerlink.4094
    clagd-args --backupPort 5400
    clagd-args --lacpPoll 900
    clagd-peer-ip linklocal
    clagd-backup-ip 192.0.2.50
    clagd-priority 8192
    clagd-sys-mac 44:38:39:ff:00:01
...

Run ifreload -a to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

In this scenario, the spine switches connect at layer 3, as shown in the image below. Alternatively, the spine switches can be singly connected to each core switch at layer 3 (not shown below).

In this design, the spine switches route traffic between the server hosts in the layer 2 domains and the core. The servers (host1 thru host4) each have a layer 2 connection up to the spine layer where the default gateway for the host subnets resides. However, since the spine switches as gateway devices communicate at layer 3, you need to configure a protocol such as VRR (virtual router redundancy) between the spine switch pair to support active/active forwarding.

Then, to connect the spine switches to the core switches, you need to determine whether the routing is static or dynamic. If it is dynamic, you must choose which protocol to use (OSPF or BGP). When enabling a routing protocol in an MLAG environment, it is also necessary to manage the uplinks, because by default MLAG is not aware of layer 3 uplink interfaces. If there is a peer link failure, MLAG does not remove static routes or bring down a BGP or OSPF adjacency unless you use a separate link state daemon such as ifplugd.

When using MLAG with VRR, Cumulus Networks recommends you set up a routed adjacency across the peerlink.4094 interface. If a routed connection is not built across the peer link, then during uplink failure on one of the switches in the MLAG pair, egress traffic can be blackholed if it hashes to the leaf whose uplinks are down.

To set up the adjacency, configure a BGP or OSPF unnumbered peering, as appropriate for your network.

For example, if you are using BGP, use a configuration like this:

cumulus@switch:~$ net add bgp neighbor peerlink.4094 interface remote-as internal
cumulus@switch:~$ net commit

If you are using OSPF, use a configuration like this:

cumulus@switch:~$ net add interface peerlink.4094 ospf area 0.0.0.1
cumulus@switch:~$ net commit

If you are using EVPN and MLAG, you need to enable the EVPN address family across the peerlink.4094 interface as well:

cumulus@switch:~$ net add bgp neighbor peerlink.4094 interface remote-as internal
cumulus@switch:~$ net add bgp l2vpn evpn neighbor peerlink.4094 activate
cumulus@switch:~$ net commit

Be aware of an existing issue when you use NCLU to create an iBGP peering, it creates an eBGP peering instead. For more information, see release note 1222.

IGMP Snooping with MLAG

IGMP snooping processes IGMP reports received on a bridge port in a bridge to identify hosts that are configured to receive multicast traffic destined to that group. An IGMP query message received on a port is used to identify the port that is connected to a router and configured to receive multicast traffic.

IGMP snooping is enabled by default on the bridge. IGMP snooping multicast database entries and router port entries are synced to the peer MLAG switch. If there is no multicast router in the VLAN, you can configure the IGMP querier on the switch to generate IGMP query messages. For more information, read the IGMP snooping chapter.

In an MLAG configuration, the switch in the secondary role does not send IGMP queries, even though the configuration is identical to the switch in the primary role. This is expected behavior, as there can be only one querier on each VLAN. Once the querier on the primary switch stops transmitting, the secondary switch starts transmitting.

Monitor the Status of the clagd Service

Due to the critical nature of the clagd service, systemd continuously monitors the status of clagd. systemd monitors the clagd service through the use of notify messages every 30 seconds. If the clagd service dies or becomes unresponsive for any reason and systemd receives no messages after 60 seconds systemd restarts clagd. systemd logs these failures in /var/log/syslog, and, on the first failure, generates a cl-supportfile as well.

This monitoring is configured and enabled automatically as long as clagd is enabled (clagd-peer-ip and clagd-sys-mac are configured for an interface) and the clagd service is running. If you stop clagd, for example with the systemctl stop clagd.service command, clagd monitoring also stops.

You can check the status of clagd monitoring by using the cl-service-summary command:

cumulus@switch:~$ sudo cl-service-summary
The systemctl daemon 5.4 uptime: 15m
...
Service clagd        enabled    active 
...

Or the systemctl status command:

cumulus@switch:~$ sudo systemctl status clagd.service
 ● clagd.service - Cumulus Linux Multi-Chassis LACP Bonding Daemon
    Loaded: loaded (/lib/systemd/system/clagd.service; enabled)
    Active: active (running) since Mon 2016-10-03 20:31:50 UTC; 4 days ago
        Docs: man:clagd(8)
    Main PID: 1235 (clagd)
    CGroup: /system.slice/clagd.service
            ├─1235 /usr/bin/python /usr/sbin/clagd --daemon 169.254.255.2 peerlink.4094 44:38:39:FF:40:90 --prior...
            └─1307 /sbin/bridge monitor fdb

Feb 01 23:19:30 leaf01 clagd[1717]: Cleanup is executing.
Feb 01 23:19:31 leaf01 clagd[1717]: Cleanup is finished
Feb 01 23:19:31 leaf01 clagd[1717]: Beginning execution of clagd version 1.3.0
Feb 01 23:19:31 leaf01 clagd[1717]: Invoked with: /usr/sbin/clagd --daemon 169.254.255.2 peerlink.4094 44:38:39:FF:40:94 --pri...168.0.12
Feb 01 23:19:31 leaf01 clagd[1717]: Role is now secondary
Feb 01 23:19:31 leaf01 clagd[1717]: Initial config loaded
Feb 01 23:19:31 leaf01 systemd[1]: Started Cumulus Linux Multi-Chassis LACP Bonding Daemon.
Feb 01 23:24:31 leaf01 clagd[1717]: HealthCheck: reload timeout.
Feb 01 23:24:31 leaf01 clagd[1717]: Role is now primary; Reload timeout
Hint: Some lines were ellipsized, use -l to show in full.

MLAG Best Practices

For MLAG to function properly, you must configure the dual-connected host interfaces identically on the pair of peering switches. See the note above in the Configure MLAG section.

MTU in an MLAG Configuration

The MTU in MLAG traffic is determined by the bridge MTU. Bridge MTU is determined by the lowest MTU setting of an interface that is a member of the bridge. If you want to set an MTU other than the default (1500 bytes on a Broadcom switch or 9238 bytes on a Mellanox switch), you must configure the MTU on each physical interface and bond interface that are members of the MLAG bridges in the entire bridged domain.

For example, if an MTU of 9216 is desired through the MLAG domain in the example shown above, on all four leaf switches, configure mtu 9216 for each of the following bond interfaces, as they are members of bridge bridge: peerlink, uplink, server01.

NCLU Commands
cumulus@switch:~$ net add bond peerlink mtu 9216
cumulus@switch:~$ net add bond uplink mtu 9216
cumulus@switch:~$ net add bond server01 mtu 9216
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. This is an example configuration:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports peerlink uplink server01

auto peerlink
iface peerlink
    mtu 9216

auto server01
iface server01
    mtu 9216

auto uplink
iface uplink
    mtu 9216
...

Run ifreload -a to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

The peer link carries very little traffic when compared to the bandwidth consumed by dataplane traffic. In a typical MLAG configuration, most every connection between the two switches in the MLAG pair is dual-connected so the only traffic going across the peer link is traffic from the clagd process and some LLDP or LACP traffic; the traffic received on the peer link is not forwarded out of the dual-connected bonds.

However, there are some instances where a host is connected to only one switch in the MLAG pair; for example:

Cumulus Networks recommends you determine how much bandwidth is traveling across the single-connected interfaces and allocate half of that bandwidth to the peer link. On average, one half of the traffic destined to the single-connected host arrives on the switch directly connected to the single-connected host and the other half arrives on the switch that is not directly connected to the single-connected host. When this happens, only the traffic that arrives on the switch that is not directly connected to the single-connected host needs to traverse the peer link.

In addition, you might want to add extra links to the peer link bond to handle link failures in the peer link bond itself.

In the illustration below, each host has two 10G links, with each 10G link going to each switch in the MLAG pair. Each host has 20G of dual-connected bandwidth, so all three hosts have a total of 60G of dual-connected bandwidth. Cumulus Networks recommend you allocate at least 15G of bandwidth to each peer link bond, which represents half of the single-connected bandwidth.

Scaling this example out to a full rack, when planning for link failures, you need only allocate enough bandwidth to meet your site’s strategy for handling failure scenarios. Imagine a full rack with 40 servers and two switches. You might plan for four to six servers to lose connectivity to a single switch and become single connected before you respond to the event. So expanding upon our previous example, if you have 40 hosts each with 20G of bandwidth dual-connected to the MLAG pair, you might allocate 20G to 30G of bandwidth to the peer link - which accounts for half of the single-connected bandwidth for four to six hosts.

Failover Redundancy Scenarios

To get a better understanding of how STP and LACP behave in response to various failover redundancy scenarios, read this knowledge base article.

STP Interoperability with MLAG

Cumulus Networks recommends that you always enable STP in your layer 2 network.

With MLAG, Cumulus Networks recommends you enable BPDU guard on the host-facing bond interfaces. For more information about BPDU guard, see BPDU Guard and Bridge Assurance.

To show useful troubleshooting information:

NCLU Commands

Run the net show bridge spanning-tree command:

cumulus@switch:~$ net show bridge spanning-tree
Bridge info
    enabled         yes
    bridge id       8.000.44:39:39:FF:40:94
    Priority:   32768
    Address:    44:39:39:FF:40:94
    This bridge is root.

designated root 8.000.44:39:39:FF:40:94
Priority:   32768
Address:    44:39:39:FF:40:94

root port       none
path cost     0          internal path cost   0
max age       20         bridge max age       20
forward delay 15         bridge forward delay 15
tx hold count 6          max hops             20
hello time    2          ageing time          300
force protocol version     rstp

E bond01 8.001 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 8.001 Desg
E bond02 8.002 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 8.002 DesgE peerlink F.003 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 F.003 Desg
E vni13 8.004 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 8.004 Desg
E vni24 8.005 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 8.005 Desg
E vxlan4001 8.006 forw 8.000.44:39:39:FF:40:94 8.000.44:39:39:FF:40:94 8.006 Desg
Linux Commands

Run the mstpctl showportdetail command:

cumulus@switch:~$ sudo mstpctl showportdetail bridge peerlink

bridge:peerlink CIST info
    enabled            yes                     role                 Designated
    port id            F.003                   state                forwarding
    external port cost 10000                   admin external cost  0
    internal port cost 10000                   admin internal cost  0
    designated root    8.000.44:39:39:FF:40:94 dsgn external cost   0
    dsgn regional root 8.000.44:39:39:FF:40:94 dsgn internal cost   0
    designated bridge  8.000.44:39:39:FF:40:94 designated port      F.003
    admin edge port    no                      auto edge port       yes
    oper edge port     yes                     topology change ack  no
    point-to-point     yes                     admin point-to-point auto
    restricted role    no                      restricted TCN       no
    port hello time    2                       disputed             no
    bpdu guard port    no                      bpdu guard error     no
    network port       no                      BA inconsistent      no
    Num TX BPDU        6                       Num TX TCN           0
    Num RX BPDU        0                       Num RX TCN           0
    Num Transition FWD 2                       Num Transition BLK   1
    bpdufilter port    no
    clag ISL           yes                     clag ISL Oper UP     yes
    clag role          primary                 clag dual conn mac   00:00:00:00:00:00
    clag remote portID F.FFF                   clag system mac      44:39:39:FF:40:94

Best Practices for STP with MLAG

  • The STP global configuration must be the same on both peer switches.
  • The STP configuration for dual-connected ports must be the same on both peer switches.
  • The STP priority must be the same on both peer switches.

For additional information on STP, see Spanning Tree Priority.

Troubleshooting

Viewing the MLAG Log File

By default, when clagd is running, it logs its status to the /var/log/clagd.log file and syslog. Example log file output is below:

cumulus@spine01:~$ sudo tail /var/log/clagd.log 
2016-10-03T20:31:50.471400+00:00 spine01 clagd[1235]: Initial config loaded
2016-10-03T20:31:52.479769+00:00 spine01 clagd[1235]: The peer switch is active.
2016-10-03T20:31:52.496490+00:00 spine01 clagd[1235]: Initial data sync to peer done.
2016-10-03T20:31:52.540186+00:00 spine01 clagd[1235]: Role is now primary; elected
2016-10-03T20:31:54.250572+00:00 spine01 clagd[1235]: HealthCheck: role via backup is primary
2016-10-03T20:31:54.252642+00:00 spine01 clagd[1235]: HealthCheck: backup active
2016-10-03T20:31:54.537967+00:00 spine01 clagd[1235]: Initial data sync from peer done.
2016-10-03T20:31:54.538435+00:00 spine01 clagd[1235]: Initial handshake done.
2016-10-03T20:31:58.527464+00:00 spine01 clagd[1235]: leaf03-04 is now dual connected.
2016-10-03T22:47:35.255317+00:00 spine01 clagd[1235]: leaf01-02 is now dual connected.

A large volume of packet drops across one of the peer link interfaces can be expected. These drops serve to prevent looping of BUM (broadcast, unknown unicast, multicast) packets. When a packet is received across the peer link, if the destination lookup results in an egress interface that is a dual-connected bond, the switch does not forward the packet to prevent loops. This results in a drop being recorded on the peer link.

You can detect this issue by running the the following commands:

NCLU Commands

Run the net show counters command. The number of dropped packets is displayed in the RX_DRP column.

cumulus@switch:~$ net show counters

Kernel Interface table
Iface              MTU    Met    RX_OK    RX_ERR    RX_DRP    RX_OVR    TX_OK    TX_ERR    TX_DRP    TX_OVR  Flg
---------------  -----  -----    -------  --------  --------  --------  -------  --------  --------  ------  -----
peerlink        1500       0      19226721     0      2952460  0       55115330     0       364      0       BMmRU
peerlink.4094   1500       0      0            0      0        0       5379243      0       0        0       BMRU
swp51           1500       0      6587220      0      2129676  0       38957769     0       202      0       BMsRU
swp52           1500       0      12639501     0      822784   0       16157561     0       162      0       BMsRU
Linux Commands

Run the ethtool -S <interface> command. The number of dropped packets are indicated by the HwIfInDiscards counter.

cumulus@switch:~$ sudo ethtool -S swp51
NIC statistics:
HwIfInOctets: 669507330
HwIfInUcastPkts: 658871
HwIfInBcastPkts: 2231559
HwIfInMcastPkts: 3696790
HwIfOutOctets: 2752224343
HwIfOutUcastPkts: 1001632
HwIfOutMcastPkts: 3743199
HwIfOutBcastPkts: 34212938
HwIfInDiscards: 2129675

Duplicate LACP Partner MAC Warning

When you run clagctl, you may see output like this:

bond01 bond01 52 duplicate lacp - partner mac

This occurs when you have multiple LACP bonds between the same two LACP endpoints; for example, an MLAG switch pair is one endpoint and an ESXi host is another. These bonds have duplicate LACP identifiers, which are MAC addresses. This same warning might trigger when you have a cabling or configuration error.

Caveats and Errata

If both backup and peer connectivity are lost within a 30-second window, the switch in the secondary role misinterprets the event sequence, sees the peer switch as down and takes over as the primary.

LACP Bypass

On Cumulus Linux, LACP Bypass allows a bond configured in 802.3ad mode to become active and forward traffic even when there is no LACP partner. For example, you can enable a host that does not have the capability to run LACP to PXE boot while connected to a switch on a bond configured in 802.3ad mode. After the pre-boot process completes and the host is capable of running LACP, the normal 802.3ad link aggregation operation takes over.

LACP Bypass All-active Mode

In all-active mode, when a bond has multiple slave interfaces, each bond slave interface operates as an active link while the bond is in bypass mode. This is useful during PXE boot of a server with multiple NICs, when you cannot determine beforehand which port needs to be active.

  • All-active mode is not supported on bonds that are not specified as bridge ports on the switch.
  • STP does not run on the individual bond slave interfaces when the LACP bond is in all-active mode. Only use all-active mode on host-facing LACP bonds. Cumulus Networks highly recommends you configure STP BPDU guard together with all-active mode.
  • In an MLAG deployment where bond slaves of a host are connected to two switches and the bond is in all-active mode, all the slaves of bond are active on both the primary and secondary MLAG nodes.
  • priority mode, bond-lacp-bypass-period, bond-lacp-bypass-priority, and bond-lacp-bypass-all-active are not supported.

Configure LACP Bypass

To enable LACP bypass on the host-facing bond, set bond-lacp-bypass-allow to yes.

NCLU Commands

The following commands create a VLAN-aware bridge with LACP bypass enabled:

cumulus@switch:~$ net add bond bond1 bond slaves swp51s2,swp51s3
cumulus@switch:~$ net add bond bond1 clag id 1
cumulus@switch:~$ net add bond bond1 bond lacp-bypass-allow
cumulus@switch:~$ net add bond bond1 stp bpduguard
cumulus@switch:~$ net add bridge bridge ports bond1,bond2,bond3,bond4,peer5
cumulus@switch:~$ net add bridge bridge vids 100-105
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file to add the set bond-lacp-bypass-allow to yes option. The following configuration creates a VLAN-aware bridge with LACP bypass enabled:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bond1
iface bond1
    bond-lacp-bypass-allow yes
    bond-slaves swp51s2 swp51s3
    clag-id 1
    mstpctl-bpduguard yes
...
auto bridge
iface bridge
    bridge-ports bond1 bond2 bond3 bond4 peer5
    bridge-vids 100-105
    bridge-vlan-aware yes
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

To check the status of the configuration, run the following commands.

NCLU Commands

Run the net show interface <bond> command on the bond and its slave interfaces:

cumulus@switch:~$ net show interface bond1

    Name   MAC               Speed   MTU   Mode
--  ------ ----------------- ------- ----- ----------
UP  bond1  44:38:39:00:00:5b 1G      1500  Bond/Trunk

Bond Details
------------------ -------------------------
Bond Mode:         LACP
Load Balancing:    Layer3+4
Minimum Links:     1
In CLAG:           CLAG Active
LACP Sys Priority:
LACP Rate:         Fast Timeout
LACP Bypass:       LACP Bypass Not Supported

    Port       Speed     TX   RX   Err   Link Failures
-- --------   ------- ---- ---- ----- ---------------
UP swp51s2(P) 1G         0    0     0               0
UP swp51s3(P) 1G         0    0     0               0


All VLANs on L2 Port
----------------------
100-105

Untagged
----------
1

Vlans in disabled State
-------------------------
100-105

LLDP
--------   ---- ------------------
swp51s2(P) ==== swp1(spine01)
swp51s3(P) ==== swp1(spine02)
Linux Commands

Run the ip link show command on the bond and its slave interfaces:

cumulus@switch:~$ ip link show bond1
164: bond1: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue master br0 state UP mode DORMANT group default 
    link/ether c4:54:44:f6:44:5a brd ff:ff:ff:ff:ff:ff
cumulus@switch:~$ ip link show swp51s2
55: swp51s2: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT group default qlen 1000
    link/ether c4:54:44:f6:44:5a brd ff:ff:ff:ff:ff:ff
cumulus@switch:~$ ip link show swp52s3
56: swp51s3: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bond1 state UP mode DEFAULT group default qlen 1000
    link/ether c4:54:44:f6:44:5a brd ff:ff:ff:ff:ff:ff

To verify that LACP bypass is enabled on a bond and its slave interfaces, use the cat command:

cumulus@switch:~$ cat /sys/class/net/bond1/bonding/lacp_bypass
on 1
cumulus@switch:~$ cat /sys/class/net/bond1/bonding/slaves
swp51 swp52
cumulus@switch:~$ cat /sys/class/net/swp52/bonding_slave/ad_rx_bypass
1
cumulus@switch:~$ cat /sys/class/net/swp51/bonding_slave/ad_rx_bypass
1

Example LACP Bypass Configuration (Traditional Bridge Mode)

The following configuration shows LACP bypass enabled for multiple active interfaces (all-active mode) with a bridge in traditional bridge mode:

...
auto bond1
iface bond1 
    bond-slaves swp3 swp4
    bond-lacp-bypass-allow 1

auto br0
iface br0
    bridge-ports bond1 bond2 bond3 bond4 peer5
    mstpctl-bpduguard bond1=yes
...

Virtual Router Redundancy - VRR and VRRP

Cumulus Linux provides the option of using Virtual Router Redundancy (VRR) or Virtual Router Redundancy Protocol (VRRP).

VRR enables hosts to communicate with any redundant router without reconfiguration, running dynamic router protocols, or running router redundancy protocols. This means that redundant routers respond to Address Resolution Protocol (ARP) requests from hosts. Routers are configured to respond in an identical manner, but if one fails, the other redundant routers continue to respond, leaving the hosts with the impression that nothing has changed. VRR is typically used in an MLAG configuration.

VRRP allows a single virtual default gateway to be shared between two or more network devices in an active/standby configuration. The physical VRRP router that forwards packets at any given time is called the master. If this VRRP router fails, another VRRP standby router automatically takes over as master. VRRP is used in a non-MLAG configuration.

You cannot configure both VRR and VRRP on the same switch.

VRR

The diagram below illustrates a basic VRR-enabled network configuration. The network includes several hosts and two routers running Cumulus Linux configured with multi-chassis link aggregation (MLAG).

Cumulus Linux only supports VRR on switched virtual interfaces (SVIs). VRR is not supported on physical interfaces or virtual subinterfaces.

A production implementation has many more server hosts and network connections than are shown here. However, this basic configuration provides a complete description of the important aspects of the VRR setup.

As the bridges in each of the redundant routers are connected, they will each receive and reply to ARP requests for the virtual router IP address.

Each ARP request made by a host will receive replies from each router; these replies will be identical, and so the host receiving the replies will either ignore replies after the first, or accept them and overwrite the previous identical reply, rather than being confused over which response is correct.

A range of MAC addresses is reserved for use with VRR to prevent MAC address conflicts with other interfaces in the same bridged network. The reserved range is 00:00:5E:00:01:00 to 00:00:5E:00:01:ff.

Cumulus Networks recommends using MAC addresses from the reserved range when configuring VRR.

The reserved MAC address range for VRR is the same as for the Virtual Router Redundancy Protocol (VRRP), as they serve similar purposes.

Configure the Routers

The routers implement the layer 2 network interconnecting the hosts and the redundant routers. To configure the routers, add a bridge with the following interfaces to each router:

NCLU Commands

The example NCLU commands below create a VLAN-aware bridge interface for a VRR-enabled network:

cumulus@switch:~$ net add bridge
cumulus@switch:~$ net add vlan 500 ip address 192.0.2.252/24
cumulus@switch:~$ net add vlan 500 ip address-virtual 00:00:5e:00:01:00 192.0.2.254/24
cumulus@switch:~$ net add vlan 500 ipv6 address 2001:db8::1/32
cumulus@switch:~$ net add vlan 500 ipv6 address-virtual 00:00:5e:00:01:00 2001:db8::f/32
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. The example file configuration below create a VLAN-aware bridge interface for a VRR-enabled network:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-vids 500
    bridge-vlan-aware yes

auto vlan500
iface vlan500
    address 192.0.2.252/24
    address 2001:db8::1/32
    address-virtual 00:00:5e:00:01:00 2001:db8::f/32 192.0.2.254/24
    vlan-id 500
    vlan-raw-device bridge
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

Configure the Hosts

Each host must have two network interfaces. The routers configure the interfaces as bonds running LACP; the hosts must also configure its two interfaces using teaming, port aggregation, port group, or EtherChannel running LACP. Configure the hosts either statically or with DHCP, with a gateway address that is the IP address of the virtual router; this default gateway address never changes.

Configure the links between the hosts and the routers in active-active mode for First Hop Redundancy Protocol.

Example VRR Configuration with MLAG

To create an MLAG configuration that incorporates VRR, use a configuration like the following:

leaf01
cumulus@leaf01:~$ net add interface eth0 ip address 192.168.0.21
cumulus@leaf01:~$ net add bond server01 bond slaves swp1-2
cumulus@leaf01:~$ net add bond server01 clag id 1
cumulus@leaf01:~$ net add bond server01 mtu 9216
cumulus@leaf01:~$ net add bond server01 alias LACP etherchannel to uplink on server01
cumulus@leaf01:~$ net add bond peerlink bond slaves swp49-50
cumulus@leaf01:~$ net add interface peerlink.4094 peerlink.4094
cumulus@leaf01:~$ net add interface peerlink.4094 ip address 169.254.255.1/30
cumulus@leaf01:~$ net add interface peerlink.4094 clag peer-ip 169.254.255.2
cumulus@leaf01:~$ net add interface peerlink.4094 clag backup-ip 192.168.0.22
cumulus@leaf01:~$ net add interface peerlink.4094 clag sys-mac 44:38:39:FF:40:90
cumulus@leaf01:~$ net add bridge bridge ports server01,peerlink
cumulus@leaf01:~$ net add bridge stp treeprio 4096
cumulus@leaf01:~$ net add vlan 100 ip address 10.0.1.2/24
cumulus@leaf01:~$ net add vlan 100 ip address-virtual 00:00:5E:00:01:01 10.0.1.1/24
cumulus@leaf01:~$ net add vlan 200 ip address 10.0.2.2/24
cumulus@leaf01:~$ net add vlan 200 ip address-virtual 00:00:5E:00:01:02 10.0.2.1/24
cumulus@leaf01:~$ net add vlan 300 ip address 10.0.3.2/24
cumulus@leaf01:~$ net add vlan 300 ip address-virtual 00:00:5E:00:01:03 10.0.3.1/24
cumulus@leaf01:~$ net add vlan 400 ip address 10.0.4.2/24
cumulus@leaf01:~$ net add vlan 400 ip address-virtual 00:00:5E:00:01:04 10.0.4.1/24
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto eth0
iface eth0
    address 192.168.0.21

auto bridge
iface bridge
    bridge-ports server01 peerlink
    bridge-vids 100 200 300 400
    bridge-vlan-aware yes
    mstpctl-treeprio 4096

auto server01
iface server01
    alias LACP etherchannel to uplink on server01
    bond-slaves swp1 swp2
    clag-id 1
    mtu 9216

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    address 169.254.255.1/30
    clagd-backup-ip 192.168.0.22
    clagd-peer-ip 169.254.255.2
    clagd-sys-mac 44:38:39:FF:40:90

auto vlan100
iface vlan100
    address 10.0.1.2/24
    address-virtual 00:00:5E:00:01:01 10.0.1.1/24
    vlan-id 100
    vlan-raw-device bridge

auto vlan200
iface vlan200
    address 10.0.2.2/24
    address-virtual 00:00:5E:00:01:02 10.0.2.1/24
    vlan-id 200
    vlan-raw-device bridge

auto vlan300
iface vlan300
    address 10.0.3.2/24
    address-virtual 00:00:5E:00:01:03 10.0.3.1/24
    vlan-id 300
    vlan-raw-device bridge

auto vlan400
iface vlan400
    address 10.0.4.2/24
    address-virtual 00:00:5E:00:01:04 10.0.4.1/24
    vlan-id 400
    vlan-raw-device bridge
leaf02
cumulus@leaf02:~$ net add interface eth0 ip address 192.168.0.22
cumulus@leaf02:~$ net add bond server01 bond slaves swp1-2
cumulus@leaf02:~$ net add bond server01 clag id 1
cumulus@leaf02:~$ net add bond server01 mtu 9216
cumulus@leaf02:~$ net add bond server01 alias LACP etherchannel to uplink on server01
cumulus@leaf02:~$ net add bond peerlink bond slaves swp49-50
cumulus@leaf02:~$ net add interface peerlink.4094 peerlink.4094
cumulus@leaf02:~$ net add interface peerlink.4094 ip address 169.254.255.2/30
cumulus@leaf02:~$ net add interface peerlink.4094 clag peer-ip 169.254.255.1
cumulus@leaf02:~$ net add interface peerlink.4094 clag backup-ip 192.168.0.21
cumulus@leaf02:~$ net add interface peerlink.4094 clag sys-mac 44:38:39:FF:40:90
cumulus@leaf02:~$ net add bridge bridge ports server01,peerlink
cumulus@leaf02:~$ net add bridge stp treeprio 4096
cumulus@leaf02:~$ net add vlan 100 ip address 10.0.1.3/24
cumulus@leaf02:~$ net add vlan 100 ip address-virtual 00:00:5E:00:01:01 10.0.1.1/24
cumulus@leaf02:~$ net add vlan 200 ip address 10.0.2.3/24
cumulus@leaf02:~$ net add vlan 200 ip address-virtual 00:00:5E:00:01:02 10.0.2.1/24
cumulus@leaf02:~$ net add vlan 300 ip address 10.0.3.3/24
cumulus@leaf02:~$ net add vlan 300 ip address-virtual 00:00:5E:00:01:03 10.0.3.1/24
cumulus@leaf02:~$ net add vlan 400 ip address 10.0.4.3/24
cumulus@leaf02:~$ net add vlan 400 ip address-virtual 00:00:5E:00:01:04 10.0.4.1/24
cumulus@leaf02:~$ net pending
cumulus@leaf02:~$ net commit

These commands create the following configuration in the /etc/network/interfaces file:

auto eth0
iface eth0
    address 192.168.0.22

auto bridge
iface bridge
    bridge-ports server01 peerlink
    bridge-vids 100 200 300 400
    bridge-vlan-aware yes
    mstpctl-treeprio 4096

auto server01
iface server01
    alias LACP etherchannel to uplink on server01
    bond-slaves swp1 swp2
    clag-id 1
    mtu 9216

auto peerlink
iface peerlink
    bond-slaves swp49 swp50

auto peerlink.4094
iface peerlink.4094
    address 169.254.255.1/30
    clagd-backup-ip 192.168.0.22
    clagd-peer-ip 169.254.255.2
    clagd-sys-mac 44:38:39:FF:40:90

auto vlan100
iface vlan100
    address 10.0.1.3/24
    address-virtual 00:00:5E:00:01:01 10.0.1.1/24
    vlan-id 100
    vlan-raw-device bridge

auto vlan200
iface vlan200
    address 10.0.2.3/24
    address-virtual 00:00:5E:00:01:02 10.0.2.1/24
    vlan-id 200
    vlan-raw-device bridge

auto vlan300
iface vlan300
    address 10.0.3.3/24
    address-virtual 00:00:5E:00:01:03 10.0.3.1/24
    vlan-id 300
    vlan-raw-device bridge

auto vlan400
iface vlan400
    address 10.0.4.3/24
    address-virtual 00:00:5E:00:01:04 10.0.4.1/24
    vlan-id 400
    vlan-raw-device bridge
server01

Create a configuration like the following on an Ubuntu host:

auto eth0
iface eth0 inet dhcp

auto eth1
iface eth1 inet manual
    bond-master uplink

auto eth2
iface eth2 inet manual
    bond-master uplink

auto uplink
iface uplink inet static
    bond-slaves eth1 eth2
    bond-mode 802.3ad
    bond-miimon 100
    bond-lacp-rate 1
    bond-min-links 1
    bond-xmit-hash-policy layer3+4
    address 172.16.1.101
    netmask 255.255.255.0
    post-up ip route add 172.16.0.0/16 via 172.16.1.1
    post-up ip route add 10.0.0.0/8 via 172.16.1.1

auto uplink:200
iface uplink:200 inet static
    address 10.0.2.101

auto uplink:300
iface uplink:300 inet static
    address 10.0.3.101

auto uplink:400
iface uplink:400 inet static
    address 10.0.4.101

# modprobe bonding
server02

Create a configuration like the following on an Ubuntu host:

auto eth0
iface eth0 inet dhcp

auto eth1
iface eth1 inet manual
    bond-master uplink

auto eth2
iface eth2 inet manual
    bond-master uplink

auto uplink
iface uplink inet static
    bond-slaves eth1 eth2
    bond-mode 802.3ad
    bond-miimon 100
    bond-lacp-rate 1
    bond-min-links 1
    bond-xmit-hash-policy layer3+4
    address 172.16.1.101
    netmask 255.255.255.0
    post-up ip route add 172.16.0.0/16 via 172.16.1.1
    post-up ip route add 10.0.0.0/8 via 172.16.1.1

auto uplink:200
iface uplink:200 inet static
    address 10.0.2.101

auto uplink:300
iface uplink:300 inet static
    address 10.0.3.101

auto uplink:400
iface uplink:400 inet static
    address 10.0.4.101

# modprobe bonding

VRRP

VRRP allows for a single virtual default gateway to be shared among two or more network devices in an active standby configuration. The VRRP router that forwards packets at any given time is called the master. If this VRRP router fails, another VRRP standby router automatically takes over as master. The master sends VRRP advertisements to other VRRP routers in the same virtual router group, which include the priority and state of the master. VRRP router priority determines the role that each virtual router plays and who becomes the new master if the master fails.

All virtual routers use 00:00:5E:00:01:XX for IPv4 gateways or 00:00:5E:00:02:XX for IPv6 gateways as their MAC address. The last byte of the address is the Virtual Router IDentifier (VRID), which is different for each virtual router in the network. This MAC address is used by only one physical router at a time, which replies with this address when ARP requests or neighbor solicitation packets are sent for the IP addresses of the virtual router.

  • Cumulus Linux supports both VRRPv2 and VRRPv3. The default protocol version is VRRPv3.
  • 255 virtual routers are supported per switch.
  • VRRP is not supported currently in an MLAG environment or with EVPN.
  • To configure VRRP on an SVI, you need to edit the /etc/frr/frr.conf file; The NCLU commands are not supported for SVIs.

RFC 5798 describes VRRP in detail.

The following example illustrates a basic VRRP configuration.

Configure VRRP

To configure VRRP, you need to specify the following information on each switch:

You can also set these optional parameters. If you do not set these parameters, the defaults are used:

Optional ParameterDefault ValueDescription
priority100The priority level of the virtual router within the virtual router group, which determines the role that each virtual router plays and what happens if the master fails. Virtual routers have a priority between 1 and 254; the router with the highest priority becomes the master.
advertisement interval1000 millisecondsThe advertisement interval is the interval between successive advertisements by the master in a virtual router group. You can specify a value between 10 and 40950.
preemptenabledPreempt mode lets the router take over as master for a virtual router group if it has a higher priority than the current master. Preempt mode is enabled by default. To disable preempt mode, you need to edit the /etc/frr/frr.conf file and add the line no vrrp <VRID> preempt to the interface stanza, then restart FRR service.

The NCLU commands write VRRP configuration to the /etc/network/interfaces file and the /etc/frr/frr.conf file.

The following example commands configure two switches (spine01 and spine02) that form one virtual router group (VRID 44) with IPv4 address 10.0.0.1/24 and IPv6 address 2001:0db8::1/64. spine01 is the master; it has a priority of 254. spine02 is the backup VRRP router.

NCLU Commands

spine01

cumulus@spine01:~$ net add interface swp1 vrrp 44 10.0.0.1/24
cumulus@spine01:~$ net add interface swp1 vrrp 44 2001:0db8::1/64
cumulus@spine01:~$ net add interface swp1 vrrp 44 priority 254
cumulus@spine01:~$ net add interface swp1 vrrp 44 advertisement-interval 5000
cumulus@spine01:~$ net pending
cumulus@spine01:~$ net commit

spine02

cumulus@spine02:~$ net add interface swp1 vrrp 44 10.0.0.1/24
cumulus@spine02:~$ net add interface swp1 vrrp 44 2001:0db8::1/64
cumulus@spine02:~$ net pending
cumulus@spine02:~$ net commit
Linux and vtysh Commands
  1. Enable the vrrpd daemon, then start the FRRouting service. See Configuring FRRouting.

  2. From the vtysh shell, configure VRRP.

    spine01

cumulus@spine01:~$ sudo vtysh

spine01# configure terminal
spine01(config)# interface swp1
spine01(config-if)# vrrp 44 ip 10.0.0.1
spine01(config-if)# vrrp 44 ipv6 2001:0db8::1
spine01(config-if)# vrrp 44 priority 254
spine01(config-if)# vrrp 44 advertisement-interval 5000
spine01(config-if)# end
spine01# write memory
spine01# exit

spine02

cumulus@spine02:~$ sudo vtysh

spine02# configure terminal
spine02(config)# interface swp1
spine02(config-if)# vrrp 44 ip 10.0.0.1
spine02(config-if)# vrrp 44 ipv6 2001:0db8::1
spine02(config-if)# end
spine02# write memory
spine02# exit

The NCLU and vtysh commands save the configuration in the /etc/frr/frr.conf file. For example:

cumulus@spine01:~$ sudo cat /etc/frr/frr.conf
...
interface swp1
vrrp 44
vrrp 44 advertisement-interval 5000
vrrp 44 priority 254
vrrp 44 ip 10.0.0.1
vrrp 44 ipv6 2001:0db8::1
...

Show VRRP Configuration

To show virtual router information on a switch, run the NCLU net show vrrp <VRID> command or the vtysh show vrrp <VRID> command. For example:

cumulus@spine01:~$ net show vrrp 44
Virtual Router ID                    44
Protocol Version                     3
Autoconfigured                       No
Shutdown                             No
Interface                            swp1
 VRRP interface (v4)                 vrrp4-3-1
VRRP interface (v6)                  vrrp6-3-1
Primary IP (v4)
Primary IP (v6)                      fe80::54df:e543:5c12:7762
Virtual MAC (v4)                     00:00:5e:00:01:01
Virtual MAC (v6)                     00:00:5e:00:02:01
Status (v4)                          Master
Status (v6)                          Master
Priority                             254
Effective Priority (v4)              254
Effective Priority (v6)              254
Preempt Mode                         Yes
Accept Mode                          Yes
Advertisement Interval               5000 ms
Master Advertisement Interval (v4)   0 ms
Master Advertisement Interval (v6)   5000 ms
Advertisements Tx (v4)               17
Advertisements Tx (v6)               17
Advertisements Rx (v4)               0
Advertisements Rx (v6)               0
Gratuitous ARP Tx (v4)               1
Neigh. Adverts Tx (v6)               1
State transitions (v4)               2
State transitions (v6)               2
Skew Time (v4)                       0 ms
Skew Time (v6)                       0 ms
Master Down Interval (v4)            0 ms
Master Down Interval (v6)            0 ms
IPv4 Addresses                       1
. . . . . . . . . . . . . . . . . .  10.0.0.1
IPv6 Addresses                       1
. . . . . . . . . . . . . . . . . .  2001:0db8::1

IGMP and MLD Snooping

IGMP (Internet Group Management Protocol) and MLD (Multicast Listener Discovery) snooping are implemented in the bridge driver in the Cumulus Linux kernel and are enabled by default. IGMP snooping processes IGMP v1/v2/v3 reports received on a bridge port in a bridge to identify the hosts which would like to receive multicast traffic destined to that group.

IGMP and MLD snooping is supported over VXLAN bridges; however, this feature is not enabled by default. To enable IGMP and MLD over VXLAN, see Configure IGMP/MLD Snooping over VXLAN.

When an IGMPv2 leave message is received, a group specific query is sent to identify if there are any other hosts interested in that group, before the group is deleted.

An IGMP query message received on a port is used to identify the port that is connected to a router and is interested in receiving multicast traffic.

MLD snooping processes MLD v1/v2 reports, queries and v1 done messages for IPv6 groups. If IGMP or MLD snooping is disabled, multicast traffic gets flooded to all the bridge ports in the bridge. Similarly, in the absence of receivers in a VLAN, multicast traffic is flooded to all ports in the VLAN. The multicast group IP address is mapped to a multicast MAC address and a forwarding entry is created with a list of ports interested in receiving multicast traffic destined to that group.

Configure IGMP/MLD Snooping over VXLAN

Cumulus Linux supports IGMP/MLD snooping over VXLAN bridges, where VXLAN ports are set as router ports, on Broadcom switches.

To enable IGMP/MLD snooping over VXLAN:

NCLU Commands
cumulus@switch:~$ net add bridge mybridge mcsnoop yes
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands
cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge.100
vlan bridge.100
  bridge-igmp-querier-src 123.1.1.1

auto bridge
iface bridge
  bridge-ports swp1 swp2 swp3
  bridge-vlan-aware yes
  bridge-vids 100 200
  bridge-pvid 1
  bridge-mcquerier 1
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

Cumulus Networks recommends that you also configure IGMP/MLD querier. See Configure IGMP/MLD Querier, below.

To disable IGMP/MLD snooping over VXLAN, run the net add bridge <bridge> mcsnoop no command.

Additional Configuration for Spectrum Switches

For switches with Spectrum ASICs, the IGMP reports received over VXLAN from remote hosts are not forwarded to the kernel, which, in certain cases, might result in local receivers not responding to the IGMP query. To workaround this issue, you need to apply certain ACL rules to avoid the IGMP report packets being sent across to the hosts:

Add the following lines to the /etc/cumulus/acl/policy.d/23_acl_test.rules file (where <swp> is the port connected to the access host), then run the cl-acltool -i command:

[ebtables]
-A FORWARD -p IPv4 -o #<swp> --ip-proto igmp -j ACCEPT --ip-destination 224.0.0.0/24
-A FORWARD -p IPv4 -o #<swp> --ip-proto igmp -j DROP

Configure IGMP/MLD Querier

If no multicast router is sending queries to configure IGMP/MLD querier on the switch, you can add a configuration similar to the following in the /etc/network/interfaces file. To enable IGMP and MLD snooping for a bridge, set bridge-mcquerier to 1 in the bridge stanza. By default, the source IP address of IGMP queries is 0.0.0.0. To set the source IP address of the queries to be the bridge IP address, configure bridge-mcqifaddr 1.

For an explanation of the relevant parameters, see the ifupdown-addons-interfaces man page.

For a VLAN-aware bridge, use a configuration like the following:

...
auto bridge.100
vlan bridge.100
  bridge-igmp-querier-src 123.1.1.1

auto bridge
iface bridge
  bridge-ports swp1 swp2 swp3
  bridge-vlan-aware yes
  bridge-vids 100 200
  bridge-pvid 1
  bridge-mcquerier 1
...

For a VLAN-aware bridge, like bridge in the above example, to enable querier functionality for VLAN 100 in the bridge, set bridge-mcquerier to 1 in the bridge stanza and set bridge-igmp-querier-src to 123.1.1.1 in the bridge.100 stanza.

You can specify a range of VLANs as well. For example:

...
auto bridge.[1-200]
vlan bridge.[1-200]
  bridge-igmp-querier-src 123.1.1.1
...

For a bridge in traditional mode, use a configuration like the following:

...
auto br0
iface br0
  address 192.0.2.10/24
  bridge-ports swp1 swp2 swp3
  bridge-vlan-aware no
  bridge-mcquerier 1
  bridge-mcqifaddr 1
...

Disable IGMP and MLD Snooping

To disable IGMP and MLD snooping, set the bridge-mcsnoop value to 0.

NCLU Commands
cumulus@switch:~$ net add bridge bridge mcsnoop no
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file and set bridge-mcsnoop to 0 in the bridge stanza:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
  bridge-mcquerier 1
  bridge-mcsnoop 0
  bridge-ports swp1 swp2 swp3
  bridge-pvid 1
  bridge-vids 100 200
  bridge-vlan-aware yes
...

Run the ifreload -a command to reload the configuration:

cumulus@switch:~$ sudo ifreload -a

Troubleshooting

To show the IGMP/MLD snooping bridge state, run the brctl showstp <bridge> command:

cumulus@switch:~$ sudo brctl showstp bridge
  bridge
  bridge id              8000.7072cf8c272c
  designated root        8000.7072cf8c272c
  root port                 0                    path cost                  0
  max age                  20.00                 bridge max age            20.00
  hello time                2.00                 bridge hello time          2.00
  forward delay            15.00                 bridge forward delay      15.00
  ageing time             300.00
  hello timer               0.00                 tcn timer                  0.00
  topology change timer     0.00                 gc timer                 263.70
  hash elasticity        4096                    hash max                4096
  mc last member count      2                    mc init query count        2
  mc router                 1                    mc snooping                1
  mc last member timer      1.00                 mc membership timer      260.00
  mc querier timer        255.00                 mc query interval        125.00
  mc response interval     10.00                 mc init query interval    31.25
  mc querier                0                    mc query ifaddr            0
  flags

swp1 (1)
  port id                8001                    state                forwarding
  designated root        8000.7072cf8c272c       path cost                  2
  designated bridge      8000.7072cf8c272c       message age timer          0.00
  designated port        8001                    forward delay timer        0.00
  designated cost           0                    hold timer                 0.00
  mc router                 1                    mc fast leave              0
  flags

swp2 (2)
  port id                8002                    state                forwarding
  designated root        8000.7072cf8c272c       path cost                  2
  designated bridge      8000.7072cf8c272c       message age timer          0.00
  designated port        8002                    forward delay timer        0.00
  designated cost           0                    hold timer                 0.00
  mc router                 1                    mc fast leave              0
  flags

swp3 (3)
  port id                8003                    state                forwarding
  designated root        8000.7072cf8c272c       path cost                  2
  designated bridge      8000.7072cf8c272c       message age timer          0.00
  designated port        8003                    forward delay timer        8.98
  designated cost           0                    hold timer                 0.00
  mc router                 1                    mc fast leave              0
  flags

To show the groups and bridge port state, run the NCLU net show bridge mdb command or the Linux bridge mdb show command. To show detailed router ports and group information, run the bridge -d -s mdb show command:

cumulus@switch:~$ sudo bridge -d -s mdb show
  dev bridge port swp2 grp 234.10.10.10 temp 241.67
  dev bridge port swp1 grp 238.39.20.86 permanent 0.00
  dev bridge port swp1 grp 234.1.1.1 temp 235.43
  dev bridge port swp2 grp ff1a::9 permanent 0.00
  router ports on bridge: swp3

Network Virtualization

Cumulus Linux supports a few forms of network virtualization.

VXLAN (Virtual Extensible LAN) is a standard overlay protocol that abstracts logical virtual networks from the physical network underneath. You can deploy simple and scalable layer 3 Clos architectures while extending layer 2 segments over that layer 3 network.

VXLAN uses a VLAN-like encapsulation technique to encapsulate MAC-based layer 2 Ethernet frames within layer 3 UDP packets. Each virtual network is a VXLAN logical layer 2 segment. VXLAN scales to 16 million segments - a 24-bit VXLAN network identifier (VNI ID) in the VXLAN header - for multi-tenancy.

Hosts on a given virtual network are joined together through an overlay protocol that initiates and terminates tunnels at the edge of the multi-tenant network, typically the hypervisor vSwitch or top of rack. These edge points are the VXLAN tunnel end points (VTEP).

Cumulus Linux can initiate and terminate VTEPs in hardware and supports wire-rate VXLAN. VXLAN provides an efficient hashing scheme across the IP fabric during the encapsulation process; the source UDP port is unique, with the hash based on layer 2 through layer 4 information from the original frame. The UDP destination port is the standard port 4789.

Cumulus Linux includes the native Linux VXLAN kernel support and integrates with controller-based overlay solutions like VMware NSX and Midokura MidoNet.

VXLAN is supported only on switches in the Cumulus Linux HCL using the Broadcom Tomahawk, Trident II, Trident II+ and Trident3 chipsets, as well as the Mellanox Spectrum chipset.

VXLAN encapsulation over layer 3 subinterfaces (for example, swp3.111) or SVIs is not supported as traffic transiting through the switch may get dropped; even if the subinterface is used only for underlay traffic and does not perform VXLAN encapsulation, traffic may still get dropped. Only configure VXLAN uplinks as layer 3 interfaces without any subinterfaces (for example, swp3).

The VXLAN tunnel endpoints cannot share a common subnet; there must be at least one layer 3 hop between the VXLAN source and destination.

Caveats and Errata

Cut-through Mode and Store and Forward Switching

On switches using Broadcom Tomahawk, Trident II, Trident II+, and Trident3 ASICs, Cumulus Linux supports store and forward switching for VXLANs but does not support cut-through mode.

On switches using Mellanox Spectrum ASICs, Cumulus Linux supports cut-through mode for VXLANs but does not support store and forward switching.

MTU Size for Virtual Network Interfaces

The maximum transmission unit (MTU) size for a virtual network interface should be 50 bytes smaller than the MTU for the physical interfaces on the switch. For more information on setting MTU, read Layer 1 and Switch Port Attributes.

Layer 3 and Layer 2 VNIs Cannot Share the Same ID

A layer 3 VNI and a layer 2 VNI cannot have the same ID. If the VNI IDs are the same, the layer 2 VNI does not get created.

Ethernet Virtual Private Network - EVPN

VXLAN is the de facto technology for implementing network virtualization in the data center, enabling layer 2 segments to be extended over an IP core (the underlay). The initial definition of VXLAN (RFC 7348) did not include any control plane and relied on a flood-and-learn approach for MAC address learning.

Overview

Ethernet Virtual Private Network (EVPN) is a standards-based control plane for VXLAN defined in RFC 7432 and draft-ietf-bess-evpn-overlay that allows for building and deploying VXLANs at scale. It relies on multi-protocol BGP (MP-BGP) to exchange information and is based on BGP-MPLS IP VPNs (RFC 4364). It enables not only bridging between end systems in the same layer 2 segment but also routing between different segments (subnets). There is also inherent support for multi-tenancy. EVPN is often referred to as the means of implementing controller-less VXLAN.

The routing control plane (including EVPN) is installed as part of the FRRouting (FRR) package. For more information about FRR, refer to FRRouting Overview.

Key Features

Cumulus Linux fully supports EVPN as the control plane for VXLAN, including for both intra-subnet bridging and inter-subnet routing, and provides these key features:

The EVPN address-family is supported with both eBGP and iBGP peering. If the underlay routing is provisioned using eBGP, you can use the same eBGP session to carry EVPN routes. For example, in a typical 2-tier Clos network topology where the leaf switches are the VTEPs, if eBGP sessions are in use between the leaf and spine switches for the underlay routing, the same sessions can be used to exchange EVPN routes; the spine switches merely act as route forwarders and do not install any forwarding state as they are not VTEPs. When EVPN routes are exchanged over iBGP peering, OSPF can be used as the IGP or the next hops can also be resolved using iBGP.

For information about VXLAN routing, including platform and hardware limitations, see VXLAN Routing.

Data plane MAC learning is disabled by default on VXLAN interfaces. Do not enable MAC learning on VXLAN interfaces: EVPN is responsible for installing remote MACs.

Basic Configuration

The following sections provide the basic configuration needed to use EVPN as the control plane for VXLAN. The steps provided assume you have already configured VXLAN interfaces, attached them to a bridge, and mapped VLANs to VNIs.

In Cumulus Linux 4.0, MAC learning is disabled and ARP/ND suppression is enabled by default. This is a change from earlier Cumulus Linux releases, where MAC learning is enabled and ARP/ND suppression disabled by default. Be sure to update any configuration scripts, if necessary.

Enable EVPN between BGP Neighbors

To enable EVPN between BGP neighbors, add the address family evpn to the existing neighbor address-family activation command.

For a non-VTEP device that is merely participating in EVPN route exchange, such as a spine switch where the network deployment uses hop-by-hop eBGP or the switch is acting as an iBGP route reflector, activating the interface for the EVPN address family is the fundamental configuration needed in FRRouting.

The other BGP neighbor address family specific configurations supported for EVPN are allowas-in and route-reflector-client.

To configure an EVPN route exchange with a BGP peer, activate the peer or peer group within the EVPN address family. For example:

NCLU Commands
cumulus@switch:~$ net add bgp autonomous-system 65000
cumulus@switch:~$ net add bgp neighbor swp1 interface remote-as external
cumulus@switch:~$ net add bgp l2vpn evpn neighbor swp1 activate
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65000
switch(config-router)# neighbor swp1 interface remote-as external
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# neighbor swp1 activate
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$ 

Adjust the remote-as above to be appropriate for your environment.

The above commands create the following configuration snippet in the /etc/frr/frr.conf file.

...
router bgp 65000
  neighbor swp1 interface remote-as external
  address-family l2vpn evpn
  neighbor swp1 activate
...

The above configuration does not result in BGP knowing about the local VNIs defined on the system and advertising them to peers. This requires additional configuration, described in Advertise All VNIs, below.

FRR is not aware of any local VNIs and MACs, or hosts (neighbors) associated with those VNIs until you enable the BGP control plane for all VNIs configured on the switch by setting the advertise-all-vni option.

This configuration is only needed on leaf switches that are VTEPs. EVPN routes received from a BGP peer are accepted, even without this explicit EVPN configuration. These routes are maintained in the global EVPN routing table. However, they only become effective (imported into the per-VNI routing table and appropriate entries installed in the kernel) when the VNI corresponding to the received route is locally known.

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn advertise-all-vni
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65000
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# advertise-all-vni
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

The above commands create the following configuration snippet in the /etc/frr/frr.conf file.

...
router bgp 65000
  neighbor swp1 interface remote-as external
  address-family l2vpn evpn
  neighbor swp1 activate
  advertise-all-vni
...

Define RDs and RTs

When FRR learns about a local VNI and there is no explicit configuration for that VNI in FRR, the route distinguisher (RD), and import and export route targets (RTs) for this VNI are automatically derived. The RD uses RouterId:VNI-Index and the import and export RTs use AS:VNI. For routes that come from a layer 2 VNI (type-2 and type-3), the RD uses the vxlan-local-tunnelip from the layer 2 VNI interface instead of the RouterId (vxlan-local-tunnelip:VNI). The RD and RTs are used in the EVPN route exchange.

The RD disambiguates EVPN routes in different VNIs (as they may have the same MAC and/or IP address) while the RTs describe the VPN membership for the route. The VNI-Index used for the RD is a unique, internally-generated number for a VNI. It only has local significance; on remote switches, its only role is for route disambiguation. This number is used instead of the VNI value itself because this number has to be less than or equal to 65535. In the RT, the AS is always encoded as a 2-byte value to allow room for a large VNI. If the router has a 4-byte AS, only the lower 2 bytes are used. This ensures a unique RT for different VNIs while having the same RT for the same VNI across routers in the same AS.

For eBGP EVPN peering, the peers are in a different AS so using an automatic RT of AS:VNI does not work for route import. Therefore, the import RT is treated as *:VNI to determine which received routes are applicable to a particular VNI. This only applies when the import RT is auto-derived and not configured.

If you do not want RDs and RTs to be derived automatically, you can define them manually. The following example commands are per VNI. You must specify these commands under address-family l2vpn evpn in BGP.

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn vni 10200 rd 172.16.100.1:20
cumulus@switch:~$ net add bgp l2vpn evpn vni 10200 route-target import 65100:20
cumulus@switch:~$ net add bgp l2vpn evpn advertise-all-vni
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# vni 10200
switch(config-router-af-vni)# rd 172.16.100.1:20
switch(config-router-af-vni)# route-target import 65100:20
switch(config-router-af-vni)# exit
switch(config-router-af)# advertise-all-vni
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

These commands create the following configuration snippet in the /etc/frr/frr.conf file.

...
address-family l2vpn evpn
  advertise-all-vni
  vni 10200
   rd 172.16.100.1:20
   route-target import 65100:20
...

If you delete the RD or RT later, it reverts back to its corresponding default value.

You can configure multiple RT values. In addition, you can configure both the import and export route targets with a single command by using route-target both:

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn vni 10400 route-target import 100:400
cumulus@switch:~$ net add bgp l2vpn evpn vni 10400 route-target import 100:500
cumulus@switch:~$ net add bgp l2vpn evpn vni 10500 route-target both 65000:500
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# vni 10400
switch(config-router-af-vni)# route-target import 100:400
switch(config-router-af-vni)# route-target import 100:500
switch(config-router-af-vni)# exit
switch(config-router-af)# vni 10500
switch(config-router-af-vni)# route-target both 65000:500
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

The above commands create the following configuration snippet in the /etc/frr/frr.conf file:

...
address-family l2vpn evpn
  vni 10400
    route-target import 100:400
    route-target import 100:500
  vni 10500
    route-target import 65000:500
    route-target export 65000:500
...

Enable EVPN in an iBGP Environment with an OSPF Underlay

You can use EVPN with an OSPF or static route underlay. This is a more complex configuration than using eBGP. In this case, iBGP advertises EVPN routes directly between VTEPs and the spines are unaware of EVPN or BGP.

The leaf switches peer with each other in a full mesh within the EVPN address family without using route reflectors. The leafs generally peer to their loopback addresses, which are advertised in OSPF. The receiving VTEP imports routes into a specific VNI with a matching route target community.

NCLU Commands
cumulus@switch:~$ net add bgp autonomous-system 65020
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.2 remote-as internal
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.3 remote-as internal
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.4 remote-as internal
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.2 activate
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.3 activate
cumulus@switch:~$ net add bgp l2vpn evpn neighbor 10.1.1.4 activate
cumulus@switch:~$ net add bgp l2vpn evpn advertise-all-vni
cumulus@switch:~$ net add ospf router-id 10.1.1.1
cumulus@switch:~$ net add loopback lo ospf area 0.0.0.0
cumulus@switch:~$ net add ospf passive-interface lo
cumulus@switch:~$ net add interface swp50 ospf area 0.0.0.0
cumulus@switch:~$ net add interface swp51 ospf area 0.0.0.0
cumulus@switch:~$ net add interface swp50 ospf network point-to-point
cumulus@switch:~$ net add interface swp51 ospf network point-to-point
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65020
switch(config-router)# neighbor 10.1.1.2 remote-as internal
switch(config-router)# neighbor 10.1.1.3 remote-as internal
switch(config-router)# neighbor 10.1.1.4 remote-as internal
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# neighbor 10.1.1.2 activate
switch(config-router-af)# neighbor 10.1.1.3 activate
switch(config-router-af)# neighbor 10.1.1.4 activate
switch(config-router-af)# advertise-all-vni
switch(config-router-af)# exit
switch(config-router)# exit
switch(config)# router ospf
switch(config-router)# router-id 10.1.1.1
switch(config-router)# passive-interface lo
switch(config-router)# exit
switch(config)# interface lo
switch(config-if)# ip ospf area 0.0.0.0
switch(config-if)# exit
switch(config)# interface swp50
switch(config-if)# ip ospf area 0.0.0.0
switch(config-if)# ospf network point-to-point
switch(config-if)# exit
switch(config)# interface swp51
switch(config-if)# ip ospf area 0.0.0.0
switch(config-if)# ospf network point-to-point
switch(config-if)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

These commands create the following configuration snippet in the /etc/frr/frr.conf file.

...
interface lo
  ip ospf area 0.0.0.0
!
interface swp50
  ip ospf area 0.0.0.0
  ip ospf network point-to-point

interface swp51
  ip ospf area 0.0.0.0
  ip ospf network point-to-point
!
router bgp 65020
  neighbor 10.1.1.2 remote-as internal
  neighbor 10.1.1.3 remote-as internal
  neighbor 10.1.1.4 remote-as internal
  !
  address-family l2vpn evpn
  neighbor 10.1.1.2 activate
  neighbor 10.1.1.3 activate
  neighbor 10.1.1.4 activate
  advertise-all-vni
  exit-address-family
  !
Router ospf
  Ospf router-id 10.1.1.1
  Passive-interface lo
...

ARP and ND Suppression

ARP suppression with EVPN allows a VTEP to suppress ARP flooding over VXLAN tunnels as much as possible. A local proxy handles ARP requests received from locally attached hosts for remote hosts. ARP suppression is the implementation for IPv4; ND suppression is the implementation for IPv6.

ARP/ND suppression is enabled by default on all VNIs in Cumulus Linux to reduce flooding of ARP/ND packets over VXLAN tunnels.

In a centralized routing deployment, you must configure layer 3 interfaces even if the switch is configured only for layer 2 (you are not using VXLAN routing). To avoid unnecessary layer 3 information from being installed, Cumulus Networks recommends you configure the ip forward off or ip6 forward off options as appropriate on the VLANs. See the example configuration below.

The following examples show a configuration using two VXLANs (10100 and 10200) and two VLANs (100 and 200).

NCLU Commands
cumulus@switch:~$ net add bridge bridge ports vni100,vni200
cumulus@switch:~$ net add bridge bridge vids 100,200
cumulus@switch:~$ net add vxlan vni100 vxlan id 10100
cumulus@switch:~$ net add vxlan vni200 vxlan id 10200
cumulus@switch:~$ net add vxlan vni100 bridge access 100
cumulus@switch:~$ net add vxlan vni200 bridge access 200
cumulus@switch:~$ net add vxlan vni100 vxlan local-tunnelip 10.0.0.1
cumulus@switch:~$ net add vxlan vni200 vxlan local-tunnelip 10.0.0.1
cumulus@switch:~$ net add vlan 100 ip forward off
cumulus@switch:~$ net add vlan 100 ipv6 forward off
cumulus@switch:~$ net add vlan 200 ip forward off
cumulus@switch:~$ net add vlan 200 ipv6 forward off
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports vni100 vni200
    bridge-stp on
    bridge-vids 100 200
    bridge-vlan-aware yes

auto vlan100
iface vlan100
    ip6-forward off
    ip-forward off
    vlan-id 100
    vlan-raw-device bridge

auto vlan200
iface vlan200
    ip6-forward off
    ip-forward off
    vlan-id 200
    vlan-raw-device bridge

auto vni100
iface vni100
    bridge-access 100
    vxlan-id 10100
    vxlan-local-tunnelip 10.0.0.1

auto vni200
iface vni200
      bridge-access 200
      vxlan-id 10200
      vxlan-local-tunnelip 10.0.0.1
...

For a bridge in traditional mode, you must edit the bridge configuration in the /etc/network/interfaces file using a text editor:

cumulus@switch:~$ sudo nano /etc/network/interfaces
...
auto bridge1
iface bridge1
    bridge-ports swp3.100 swp4.100 vni100
    ip6-forward off
    ip-forward off
...

When deploying EVPN and VXLAN using a hardware profile other than the default Forwarding Table Profile, ensure that the Linux kernel ARP sysctl settings gc_thresh2 and gc_thresh3 are both set to a value larger than the number of neighbor (ARP/ND) entries anticipated in the deployment. To configure these settings, edit the /etc/sysctl.d/neigh.conf file, then reboot the switch. If your network has more hosts than the values used in the example below, change the sysctl entries accordingly.

Example /etc/sysctl.d/neigh.conf file
cumulus@switch:~$ sudo nano /etc/sysctl.d/neigh.conf
...
net.ipv4.neigh.default.gc_thresh3=14336
net.ipv6.neigh.default.gc_thresh3=16384
net.ipv4.neigh.default.gc_thresh2=7168
net.ipv6.neigh.default.gc_thresh2=8192
...

Cumulus Networks recommends that you keep ARP and ND suppression enabled to reduce flooding of ARP/ND packets over VXLAN tunnels. However, if you need to disable ARP and ND suppression, follow the example commands below.

NCLU Commands
cumulus@switch:~$ net add vxlan vni100 bridge arp-nd-suppress off
cumulus@switch:~$ net add vxlan vni200 bridge arp-nd-suppress off
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

Linux Commands

Edit the /etc/network/interfaces file.

cumulus@switch:~$ sudo nano /etc/network/interfaces
...

auto vni100
iface vni100
    bridge-access 100
    bridge-arp-nd-suppress off
    vxlan-id 10100
    vxlan-local-tunnelip 10.0.0.1

auto vni200
iface vni200
      bridge-access 200
      bridge-arp-nd-suppress off
      vxlan-id 10200
      vxlan-local-tunnelip 10.0.0.1
...

EVPN and VXLAN Active-active Mode

For EVPN in VXLAN active-active mode, both switches in the MLAG pair establish EVPN peering with other EVPN speakers (for example, with spine switches if using hop-by-hop eBGP) and inform about their locally known VNIs and MACs. When MLAG is active, both switches announce this information with the shared anycast IP address.

For active-active configuration, make sure that:

MLAG synchronizes information between the two switches in the MLAG pair; EVPN does not synchronize.

For type-5 routes in an EVPN symmetric configuration with VXLAN active-active mode, Cumulus Linux uses Primary IP Address Advertisement. For information on configuring Primary IP Address Advertisement, see Advertise Primary IP Address.

For information about active-active VTEPs and anycast IP behavior, and for failure scenarios, see VXLAN Active-Active Mode.

Caveats

EVPN Enhancements

This section describes EVPN enhancements.

Configure Static MAC Addresses

MAC addresses that are intended to be pinned to a particular VTEP can be provisioned on the VTEP as a static bridge FDB entry. EVPN picks up these MAC addresses and advertises them to peers as remote static MACs. You configure static bridge FDB entries for MACs under the bridge configuration:

NCLU Commands
cumulus@switch:~$ net add bridge post-up bridge fdb add 00:11:22:33:44:55 dev swp1 vlan 101 master static
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit

For a bridge in traditional mode, you must edit the bridge configuration in the /etc/network/interfaces file using a text editor:

cumulus@leaf01:~$ sudo nano /etc/network/interfaces
...
auto br101
iface br101
    bridge-ports swp1.101 vni10101
    post-up bridge fdb add 00:11:22:33:44:55 dev swp1.101 master static
...
Linux Commands

Edit the /etc/network/interfaces file. For example:

cumulus@leaf01:~$ sudo nano /etc/network/interfaces
...
auto bridge
iface bridge
    bridge-ports swp1 vni10101
    bridge-vids 101
    bridge-vlan-aware yes
    post-up bridge fdb add 00:11:22:33:44:55 dev swp1 vlan 101 master static
...

Filter EVPN Routes

A common deployment scenario for large data centers is to sub divide the data center into multiple pods with full host mobility within a pod but only do prefix-based routing across pods. You can achieve this by only exchanging EVPN type-5 routes across pods.

To filter EVPN routes based on the route-type and allow only certain types of EVPN routes to be advertised in the fabric:

NCLU Commands

Use the net add routing route-map <route_map_name> (deny|permit) <1-65535> match evpn default-route command or the net add routing route-map <route_map_name> (deny|permit) <1-65535> match evpn route-type (macip|prefix|multicast) command.

The following example commands configure EVPN to advertise type-5 routes only:

cumulus@switch:~$ net add routing route-map map1 permit 1 match evpn route-type prefix
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands

The following example configures EVPN to advertise type-5 routes only:

cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# route-map map1 permit 1
switch(config)# match evpn route-type prefix
switch(config)# end
switch# write memory
switch# exit
cumulus@switch:~$ 

In a typical EVPN deployment, you reuse SVI IP addresses on VTEPs across multiple racks. However, if you use unique SVI IP addresses across multiple racks and you want the local SVI IP address to be reachable via remote VTEPs, you can enable the advertise-svi-ip option. This option advertises the SVI IP/MAC address as a type-2 route and eliminates the need for any flooding over VXLAN to reach the IP from a remote VTEP/rack.

  • When you enable the advertise-svi-ip option, the anycast IP/MAC address pair is not advertised. Be sure not to enable both the advertise-svi-ip option and the advertise-default-gw option at the same time. (The advertise-default-gw option configures the gateway VTEPs to advertise their IP/MAC address. See Advertising the Default Gateway.
  • If your switch is in an MLAG configuration, refer to Advertise Primary IP Address.

To advertise all SVI IP/MAC addresses on the switch, run these commands:

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn advertise-svi-ip
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# advertise-svi-ip
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

cumulus@switch:~$ sudo cat /etc/frr/frr.conf
...
address-family l2vpn evpn
  advertise-svi-ip
exit-address-family
...

To advertise a specific SVI IP/MAC address, run these commands:

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn vni 10 advertise-svi-ip
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# vni 10
switch(config-router-af-vni)# advertise-svi-ip
switch(config-router-af-vni)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

The NCLU and vtysh commands save the configuration in the /etc/frr/frr.conf file. For example:

cumulus@switch:~$ sudo cat /etc/frr/frr.conf
...
address-family l2vpn evpn
  vni 10
  advertise-svi-ip
exit-address-family
...

Extended Mobility

Cumulus Linux supports scenarios where the IP to MAC binding for a host or virtual machine changes across the move. In addition to the simple mobility scenario where a host or virtual machine with a binding of IP1, MAC1 moves from one rack to another, Cumulus Linux supports additional scenarios where a host or virtual machine with a binding of IP1, MAC1 moves and takes on a new binding of IP2, MAC1 or IP1, MAC2. The EVPN protocol mechanism to handle extended mobility continues to use the MAC mobility extended community and is the same as the standard mobility procedures. Extended mobility defines how the sequence number in this attribute is computed when binding changes occur.

Extended mobility not only supports virtual machine moves, but also where one virtual machine shuts down and another is provisioned on a different rack that uses the IP address or the MAC address of the previous virtual machine. For example, in an EVPN deployment with OpenStack, where virtual machines for a tenant are provisioned and shut down very dynamically, a new virtual machine can use the same IP address as an earlier virtual machine but with a different MAC address.

The support for extended mobility is enabled by default and does not require any additional configuration.

You can examine the sequence numbers associated with a host or virtual machine MAC address and IP address with the NCLU net show evpn mac vni <vni> mac <address> command or the vtysh show evpn mac vni <vni> mac <address> command. For example:

cumulus@switch:~$ net show evpn mac vni 10100 mac 00:02:00:00:00:42
MAC: 00:02:00:00:00:42
  Remote VTEP: 10.0.0.2
  Local Seq: 0 Remote Seq: 3
  Neighbors:
    10.1.1.74 Active

cumulus@switch:~$ net show evpn arp vni 10100 ip 10.1.1.74
IP: 10.1.1.74
  Type: local
  State: active
  MAC: 44:39:39:ff:00:24
  Local Seq: 2 Remote Seq: 3

Duplicate Address Detection

Cumulus Linux is able to detect duplicate MAC and IPv4/IPv6 addresses on hosts or virtual machines in a VXLAN-EVPN configuration. The Cumulus Linux switch (VTEP) considers a host MAC or IP address to be duplicate if the address moves across the network more than a certain number of times within a certain number of seconds (five moves within 180 seconds by default). In addition to legitimate host or VM mobility scenarios, address movement can occur when IP addresses are misconfigured on hosts or when packet looping occurs in the network due to faulty configuration or behavior.

Duplicate address detection is enabled by default and triggers when:

By default, when a duplicate address is detected, Cumulus Linux flags the address as a duplicate and generates an error in syslog so that you can troubleshoot the reason and address the fault, then clear the duplicate address flag. No functional action is taken on the address.

If a MAC address is flagged as a duplicate, all IP addresses associated with that MAC are flagged as duplicates.

In an MLAG configuration, MAC mobility detection runs independently on each switch in the MLAG pair. Based on the sequence in which local learning and/or route withdrawal from the remote VTEP occurs, a type-2 route might have its MAC mobility counter incremented only on one of the switches in the MLAG pair. In rare cases, it is possible for neither VTEP to increment the MAC mobility counter for the type-2 prefix.

When Does Duplicate Address Detection Trigger?

The VTEP that sees an address move from remote to local begins the detection process by starting a timer. Each VTEP runs duplicate address detection independently. Detection always starts with the first mobility event from remote to local. If the address is initially remote, the detection count can start with the very first move for the address. If the address is initially local, the detection count starts only with the second or higher move for the address. If an address is undergoing a mobility event between remote VTEPs, duplicate detection is not started.

The following illustration shows VTEP-A, VTEP-B, and VTEP-C in an EVPN configuration. Duplicate address detection triggers on VTEP-A when there is a duplicate MAC address for two hosts attached to VTEP-A and VTEP-B. However, duplicate detection does not trigger on VTEP-A when mobility events occur between two remote VTEPs (VTEP-B and VTEP-C).

Configure Duplicate Address Detection

To change the threshold for MAC and IP address moves, run the net add bgp l2vpn evpn dup-addr-detection max-moves <number-of-events> time <duration> command. You can specify max-moves to be between 2 and 1000 and time to be between 2 and 1800 seconds.

The following example command sets the maximum number of address moves allowed to 10 and the duplicate address detection time interval to 1200 seconds.

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn dup-addr-detection max-moves 10 time 1200 
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# dup-addr-detection max-moves 10 time 1200
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$ 

To disable duplicate address detection, see Disable Duplicate Address Detection below.

Example syslog Messages

The following example shows the syslog message that is generated when Cumulus Linux detects a MAC address as a duplicate during a local update:

2018/11/06 18:55:29.463327 ZEBRA: [EC 4043309149] VNI 1001: MAC 00:01:02:03:04:11 detected as duplicate during local update, last VTEP 172.16.0.16

The following example shows the syslog message that is generated when Cumulus Linux detects an IP address as a duplicate during a remote update:

2018/11/09 22:47:15.071381 ZEBRA: [EC 4043309151] VNI 1002: MAC aa:22:aa:aa:aa:aa IP 10.0.0.9 detected as duplicate during remote update, from VTEP 172.16.0.16

Freeze a Detected Duplicate Address

Cumulus Linux provides a freeze option that takes action on a detected duplicate address. You can freeze the address permanently (until you intervene) or for a defined amount of time, after which it is cleared automatically.

When you enable the freeze option and a duplicate address is detected:

To recover from a freeze, shut down the faulty host or VM or fix any other misconfiguration in the network. If the address is frozen permanently, issue the clear command on the VTEP where the address is marked as duplicate. If the address is frozen for a defined period of time, it is cleared automatically after the timer expires (you can clear the duplicate address before the timer expires with the clear command).

If you issue the clear command or the timer expires before you address the fault, duplicate address detection might occur repeatedly.

After you clear a frozen address, if it is present behind a remote VTEP, the kernel and hardware forwarding tables are updated. If the address is locally learned on this VTEP, the address is advertised to remote VTEPs. All VTEPs get the correct address as soon as the host communicates . Silent hosts are learned only after the faulty entries age out, or you intervene and clear the faulty MAC and ARP table entries.

Configure the Freeze Option

To enable Cumulus Linux to freeze detected duplicate addresses, run the net add bgp l2vpn evpn dup-addr-detection freeze <duration>|permanent command. The duration can be any number of seconds between 30 and 3600.

The following example command freezes duplicate addresses for a period of 1000 seconds, after which it is cleared automatically:

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn dup-addr-detection freeze 1000
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# dup-addr-detection freeze 1000
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$ 

Cumulus Networks recommends you set the freeze timer to be three times the duplicate address detection window. For example, if the duplicate address detection window is set to the default of 180 seconds, set the freeze timer to 540 seconds.

The following example command freezes duplicate addresses permanently (until you issue the clear command):

NCLU Commands
cumulus@switch:~$ net add bgp l2vpn evpn dup-addr-detection freeze permanent
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# dup-addr-detection freeze permanent
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$ 

Clear Duplicate Addresses

You can clear a duplicate MAC or IP address (and unfreeze a frozen address). The following example command clears IP address 10.0.0.9 for VNI 101.

NCLU Commands
cumulus@switch:~$ net clear evpn dup-addr vni 101 ip 10.0.0.9
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# clear evpn dup-addr vni 101 ip 10.0.0.9
switch)# exit
cumulus@switch:~$ 

To clear duplicate addresses for all VNIs, run the following command:

NCLU Commands
cumulus@switch:~$ net clear evpn dup-addr vni all
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# clear evpn dup-addr vni all
switch)# exit
cumulus@switch:~$ 

In an MLAG configuration, you need to run the clear command on both the MLAG primary and secondary switch.

When you clear a duplicate MAC address, all its associated IP addresses are also cleared. However, you cannot clear an associated IP address if its MAC address is still in a duplicate state.

Disable Duplicate Address Detection

By default, duplicate address detection is enabled and a syslog error is generated when a duplicate address is detected. To disable duplicate address detection, run the following command.

NCLU Commands
cumulus@switch:~$ net del bgp l2vpn evpn dup-addr-detection
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# no dup-addr-detection
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$

When you disable duplicate address detection, Cumulus Linux clears the configuration and all existing duplicate addresses.

Show Detected Duplicate Address Information

During the duplicate address detection process, you can see the start time and current detection count with the NCLU net show evpn mac vni <vni_id> mac <mac_addr> command or the vtysh show evpn mac vni <vni_id> mac <mac_addr> command. The following command example shows that detection started for MAC address 00:01:02:03:04:11 for VNI 1001 on Tuesday, Nov 6 at 18:55:05 and the number of moves detected is 1.

cumulus@switch:~$ net show evpn mac vni 1001 mac 00:01:02:03:04:11
MAC: 00:01:02:03:04:11
  Intf: hostbond3(15) VLAN: 1001
  Local Seq: 1 Remote Seq: 0
  Duplicate detection started at Tue Nov  6 18:55:05 2018, detection count 1
  Neighbors:
    10.0.1.26 Active

After the duplicate MAC address is cleared, the NCLU net show evpn mac vni <vni_id> mac <mac_addr> command or the vtysh show evpn mac vni <vni_id> mac <mac_addr> command shows:

MAC: 00:01:02:03:04:11
  Remote VTEP: 172.16.0.16
  Local Seq: 13 Remote Seq: 14
  Duplicate, detected at Tue Nov  6 18:55:29 2018
  Neighbors:
    10.0.1.26 Active

To display information for a duplicate IP address, run the NCLU net show evpn arp-cache vni <vni_id> ip <ip_addr> command or the vtysh show evpn arp-cache vni <vni_id> ip <ip_addr> command. The following command example shows information for IP address 10.0.0.9 for VNI 1001.

cumulus@switch:~$ net show evpn arp-cache vni 1001 ip 10.0.0.9
IP: 10.0.0.9
  Type: remote
  State: inactive
  MAC: 00:01:02:03:04:11
  Remote VTEP: 10.0.0.34
  Local Seq: 0 Remote Seq: 14
  Duplicate, detected at Tue Nov  6 18:55:29 2018

To show a list of MAC addresses detected as duplicate for a specific VNI or for all VNIs, run the NCLU net show evpn mac vni <vni-id|all> duplicate command or the vtysh show evpn mac vni <vni-id|all> duplicate command. The following example command shows a list of duplicate MAC addresses for VNI 1001:

cumulus@switch:~$ net show evpn mac vni 1001 duplicate
Number of MACs (local and remote) known for this VNI: 16
MAC               Type   Intf/Remote VTEP      VLAN
aa:bb:cc:dd:ee:ff local  hostbond3             1001  

To show a list of IP addresses detected as duplicate for a specific VNI or for all VNIs, run the NCLU net show evpn arp-cache vni <vni-id|all> duplicate command or the vtysh show evpn arp-cache vni <vni-id|all> duplicate command. The following example command shows a list of duplicate IP addresses for VNI 1001:

cumulus@switch:~$ net show evpn arp-cache vni 1001 duplicate
Number of ARPs (local and remote) known for this VNI: 20
IP                Type   State    MAC                Remote VTEP
10.0.0.8          local  active   aa:11:aa:aa:aa:aa
10.0.0.9          local  active   aa:11:aa:aa:aa:aa
10.10.0.12        remote active   aa:22:aa:aa:aa:aa  172.16.0.16

To show configured duplicate address detection parameters, run the NCLU net show evpn command or the vtysh show evpn command:

cumulus@switch:~$ net show evpn
L2 VNIs: 4
L3 VNIs: 2
Advertise gateway mac-ip: No
Duplicate address detection: Enable
  Detection max-moves 7, time 300
  Detection freeze permanent

Inter-subnet Routing

There are multiple models in EVPN for routing between different subnets (VLANs), also known as inter-VLAN routing. The model you choose depends if every VTEP acts as a layer 3 gateway and performs routing or if only specific VTEPs perform routing, and if routing is performed only at the ingress of the VXLAN tunnel or both the ingress and the egress of the VXLAN tunnel.

Cumulus Linux supports these models:

Distributed routing (asymmetric or symmetric) is commonly deployed with the VTEPs configured with an anycast IP/MAC address for each subnet; each VTEP that has a particular subnet is configured with the same IP/MAC for that subnet. Such a model facilitates easy host/VM mobility as there is no need to change the host/VM configuration when it moves from one VTEP to another.

All routing occurs in the context of a tenant VRF (virtual routing and forwarding). A VRF instance is provisioned for each tenant and the subnets of the tenant are associated with that VRF (the corresponding SVI is attached to the VRF). Inter-subnet routing for each tenant occurs within the context of the VRF for that tenant and is separate from the routing for other tenants.

Centralized Routing

In centralized routing, you configure a specific VTEP to act as the default gateway for all the hosts in a particular subnet throughout the EVPN fabric. It is common to provision a pair of VTEPs in active-active mode as the default gateway using an anycast IP/MAC address for each subnet. You need to configure all subnets on such a gateway VTEP. When a host in one subnet wants to communicate with a host in another subnet, it addresses the packets to the gateway VTEP. The ingress VTEP (to which the source host is attached) bridges the packets to the gateway VTEP over the corresponding VXLAN tunnel. The gateway VTEP performs the routing to the destination host and post-routing, the packet gets bridged to the egress VTEP (to which the destination host is attached). The egress VTEP then bridges the packet on to the destination host.

To enable centralized routing, you must configure the gateway VTEPs to advertise their IP/MAC address. Use the advertise-default-gw command:

NCLU Commands
cumulus@leaf01:~$ net add bgp autonomous-system 65000
cumulus@leaf01:~$ net add bgp l2vpn evpn advertise-default-gw
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011
switch(config-router)# address-family l2vpn evpn 
switch(config-router-af)# advertise-default-gw
switch(config-router-af)# end
switch)# write memory
switch)# exit
cumulus@switch:~$ 

These commands create the following configuration snippet in the /etc/frr/frr.conf file.

...
router bgp 65000
...
  address-family l2vpn evpn
    advertise-default-gw
  exit-address-family
...

You can deploy centralized routing at the VNI level. Therefore, you can configure the advertise-default-gw command per VNI so that centralized routing is used for some VNIs while distributed routing (described below) is used for other VNIs. This type of configuration is not recommended unless the deployment requires it.

When centralized routing is in use, even if the source host and destination host are attached to the same VTEP, the packets travel to the gateway VTEP to get routed and then come back.

Asymmetric Routing

In distributed asymmetric routing, each VTEP acts as a layer 3 gateway, performing routing for its attached hosts. The routing is called asymmetric because only the ingress VTEP performs routing, the egress VTEP only performs bridging. Asymmetric routing can be achieved with only host routing and does not involve any interconnecting VNIs. However, you must provision each VTEP with all VLANs/VNIs - the subnets between which communication can take place; this is required even if there are no locally-attached hosts for a particular VLAN.

The only additional configuration required to implement asymmetric routing beyond the standard configuration for a layer 2 VTEP described earlier is to ensure that each VTEP has all VLANs (and corresponding VNIs) provisioned on it and the SVI for each such VLAN is configured with an anycast IP/MAC address.

Symmetric Routing

In distributed symmetric routing, each VTEP acts as a layer 3 gateway, performing routing for its attached hosts; however, both the ingress VTEP and egress VTEP route the packets (similar to the traditional routing behavior of routing to a next hop router). In the VXLAN encapsulated packet, the inner destination MAC address is set to the router MAC address of the egress VTEP as an indication that the egress VTEP is the next hop and also needs to perform routing. All routing happens in the context of a tenant (VRF). For a packet received by the ingress VTEP from a locally attached host, the SVI interface corresponding to the VLAN determines the VRF. For a packet received by the egress VTEP over the VXLAN tunnel, the VNI in the packet has to specify the VRF. For symmetric routing, this is a VNI corresponding to the tenant and is different from either the source VNI or the destination VNI. This VNI is referred to as the layer 3 VNI or interconnecting VNI; it has to be provisioned by the operator and is exchanged through the EVPN control plane. To make the distinction clear, the regular VNI, which is used to map a VLAN, is referred to as the layer 2 VNI.

  • There is a one-to-one mapping between a layer 3 VNI and a tenant (VRF).
  • The VRF to layer 3 VNI mapping has to be consistent across all VTEPs. The layer 3 VNI has to be provisioned by the operator.
  • A layer 3 VNI and a layer 2 VNI cannot have the same ID. If the VNI IDs are the same, the layer 2 VNI does not get created.
  • In an MLAG configuration, the SVI used for the layer 3 VNI cannot be part of the bridge. This ensures that traffic tagged with that VLAN ID is not forwarded on the peer link or other trunks.

In an EVPN symmetric routing configuration, when a type-2 (MAC/IP) route is announced, in addition to containing two VNIs (the layer 2 VNI and the layer 3 VNI), the route also contains separate RTs for layer 2 and layer 3. The layer 3 RT associates the route with the tenant VRF. By default, this is auto-derived in a similar way to the layer 2 RT, using the layer 3 VNI instead of the layer 2 VNI; however you can also explicitly configure it.

For EVPN symmetric routing, additional configuration is required:

Optional configuration includes configuring RD and RTs for the tenant VRF and advertising the locally-attached subnets.

Configure a Per-tenant VXLAN Interface

NCLU Commands
cumulus@leaf01:~$ net add vxlan vni104001 vxlan id 104001
cumulus@leaf01:~$ net add vxlan vni104001 bridge access 4001
cumulus@leaf01:~$ net add vxlan vni104001 vxlan local-tunnelip 10.0.0.11
cumulus@leaf01:~$ net add bridge bridge ports vni104001
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. For example:

cumulus@leaf01:~$ sudo nano /etc/network/interfaces
...
auto vni104001
iface vni104001
    bridge-access 4001
    vxlan-id 104001
    vxlan-local-tunnelip 10.0.0.11

auto bridge
  iface bridge
    bridge-ports vni104001
    bridge-vlan-aware yes
...

Configure an SVI for the Layer 3 VNI

NCLU Commands
cumulus@leaf01:~$ net add vlan 4001 vrf turtle
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
Linux Commands

Edit the /etc/network/interfaces file. For example:

cumulus@leaf01:~$ sudo nano /etc/network/interfaces
...
auto vlan4001
iface vlan4001
    vlan-id 4001
    vlan-raw-device bridge
    vrf turtle
...

When two VTEPs are operating in VXLAN active-active mode and performing symmetric routing, you need to configure the router MAC corresponding to each layer 3 VNI to ensure both VTEPs use the same MAC address. Specify the address-virtual (MAC address) for the SVI corresponding to the layer 3 VNI. Use the same address on both switches in the MLAG pair. Cumulus Networks recommends you use the MLAG system MAC address. See Advertise Primary IP Address.

Configure the VRF to Layer 3 VNI Mapping

NCLU Commands
cumulus@leaf01:~$ net add vrf turtle vni 104001
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
Linux Commands

Edit the /etc/frr/frr.conf file. For example:

cumulus@leaf01:~$ sudo nano /etc/frr/frr.conf
...
vrf turtle
  vni 104001
!
...

Configure RD and RTs for the Tenant VRF

If you do not want the RD and RTs (layer 3 RTs) for the tenant VRF to be derived automatically, you can configure them manually by specifying them under the l2vpn evpn address family for that specific VRF.

NCLU Commands
cumulus@switch:~$ net add bgp vrf tenant1 l2vpn evpn rd 172.16.100.1:20
cumulus@switch:~$ net add bgp vrf tenant1 l2vpn evpn route-target import 65100:20
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011 vrf tenant1
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# rd 172.16.100.1:20
switch(config-router-af)# route-target import 65100:20
switch(config-router-af)# end
switch# write memory
switch# exit
cumulus@switch:~$

These commands create the following configuration snippet in the /etc/frr/frr.conf file:

...
router bgp <as> vrf tenant1
  address-family l2vpn evpn
  rd 172.16.100.1:20
  route-target import 65100:20
...

The tenant VRF RD and RTs are different from the RD and RTs for the layer 2 VNI. See Define RDs and RTs.

Symmetric routing presents a problem in the presence of silent hosts. If the ingress VTEP does not have the destination subnet and the host route is not advertised for the destination host, the ingress VTEP cannot route the packet to its destination. You can overcome this problem by having VTEPs announce the subnet prefixes corresponding to their connected subnets in addition to announcing host routes. These routes are announced as EVPN prefix (type-5) routes.

To advertise locally attached subnets:

  1. Enable advertisement of EVPN prefix (type-5) routes. Refer to Prefix-based Routing - EVPN Type-5 Routes, below.
  2. Ensure that the routes corresponding to the connected subnets are known in the BGP VRF routing table by injecting them using the network command or redistributing them using the redistribute connected command.

This configuration is recommended only if the deployment is known to have silent hosts. It is also recommended that you enable on only one VTEP per subnet, or two for redundancy.

Prefix-based Routing

EVPN in Cumulus Linux supports prefix-based routing using EVPN type-5 (prefix) routes. Type-5 routes (or prefix routes) are primarily used to route to destinations outside of the data center fabric.

EVPN prefix routes carry the layer 3 VNI and router MAC address and follow the symmetric routing model for routing to the destination prefix.

  • When connecting to a WAN edge router to reach destinations outside the data center, Cumulus Networks recommends that you deploy specific border/exit leaf switches to originate the type-5 routes.
  • On switches with Spectrum ASICs, centralized routing, symmetric routing, and prefix-based routing only work with the Spectrum A1 chip.
  • If you are using a Broadcom Trident II+ switch as a border/exit leaf, see the Inter-subnet Routing below for a required workaround; the workaround only applies to Trident II+ switches, not Tomahawk or Spectrum.

Install EVPN Type-5 Routes

For a switch to be able to install EVPN type-5 routes into the routing table, you must configure it with the layer 3 VNI related information. This configuration is the same as for symmetric routing. You need to:

  1. Configure a per-tenant VXLAN interface that specifies the layer 3 VNI for the tenant. This VXLAN interface is part of the bridge; router MAC addresses of remote VTEPs are installed over this interface.
  2. Configure an SVI (layer 3 interface) corresponding to the per-tenant VXLAN interface. This is attached to the VRF of the tenant. The remote prefix routes are installed over this SVI.
  3. Specify the mapping of the VRF to layer 3 VNI. This configuration is for the BGP control plane.

Announce EVPN Type-5 Routes

The following configuration is required in the tenant VRF to announce IP prefixes in the BGP RIB as EVPN type-5 routes.

NCLU Commands
cumulus@switch:~$ net add bgp vrf vrf1 l2vpn evpn advertise ipv4 unicast
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011 vrf vrf1
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# advertise ipv4 unicast
switch(config-router-af)# end
switch# write memory
switch# exit
cumulus@switch:~$

These commands create the following snippet in the /etc/frr/frr.conf file:

...
router bgp 65005 vrf vrf1
  address-family l2vpn evpn
    advertise ipv4 unicast
  exit-address-family
end
...

EVPN Type-5 Routing in Asymmetric Mode

Asymmetric routing is an ideal choice when all VLANs (subnets) are configured on all leaf switches. It simplifies the routing configuration and eliminates the potential need for advertising subnet routes to handle silent hosts. However, most deployments need access to external networks to reach the Internet or global destinations, or to do subnet-based routing between pods or data centers; this requires EVPN type-5 routes.

Cumulus Linux supports EVPN type-5 routes for prefix-based routing in asymmetric configurations within the pod or data center by providing an option to use the layer 3 VNI only for type-5 routes; type-2 routes (host routes) only use the layer 2 VNI.

The following example commands show how to use the layer 3 VNI for type-5 routes only:

NCLU Commands
cumulus@leaf01:~$ net add vrf turtle vni 104001 prefix-routes-only
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit

There is no command to delete the prefix-routes-only option. The net del vrf <vrf> vni <vni> prefix-routes-only command deletes the VNI.

Linux Commands

Edit the /etc/frr/frr.conf file. For example:

cumulus@leaf01:~$ sudo nano /etc/frr/frr.conf
...
vrf turtle
  vni 104001 prefix-routes-only
...

Control RIB Routes

By default, when announcing IP prefixes in the BGP RIB as EVPN type-5 routes, all routes in the BGP RIB are picked for advertisement as EVPN type-5 routes. You can use a route map to allow selective advertisement of routes from the BGP RIB as EVPN type-5 routes.

The following commands add a route map filter to IPv4 EVPN type-5 route advertisement:

NCLU Commands
cumulus@switch:~$ net add bgp vrf turtle l2vpn evpn advertise ipv4 unicast route-map map1
cumulus@switch:~$ net pending
cumulus@switch:~$ net commit
vtysh Commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65011 vrf turtle
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# advertise ipv4 unicast route-map map1
switch(config-router-af)# end
switch# write memory
switch# exit
cumulus@switch:~$

Originate Default EVPN Type-5 Routes

Cumulus Linux supports originating EVPN default type-5 routes. The default type-5 route is originated from a border (exit) leaf and advertised to all the other leafs within the pod. Any leaf within the pod follows the default route towards the border leaf for all external traffic (towards the Internet or a different pod).

To originate a default type-5 route in EVPN, you need to execute FRRouting commands. The following shows an example:

cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 650030 vrf vrf1
switch(config-router)# address-family l2vpn evpn
switch(config-router-af)# default-originate ipv4
switch(config-router-af)# default-originate ipv6
switch(config-router-af)# end
switch# write memory

With Cumulus Linux 3.7 and earlier, in EVPN symmetric routing configurations with VXLAN active-active (MLAG), all EVPN routes are advertised with the anycast IP address (clagd-vxlan-anycast-ip) as the next-hop IP address and the anycast MAC address as the router MAC address. In a failure scenario, this can lead to traffic being forwarded to a leaf switch that does not have the destination routes. Traffic has to traverse the peer link (with additional BGP sessions per VRF).

To prevent sub-optimal routing in Cumulus Linux 4.0 and later, the next hop IP address of the VTEP is conditionally handled depending on the route type: host type-2 (MAC/IP advertisement) or type-5 (IP prefix route).

See EVPN and VXLAN Active-Active mode for information about EVPN and VXLAN active-active mode.

Configure Advertise Primary IP Address

Run the address-virtual <anycast-mac> command under the SVI, where <anycast-mac> is the MLAG system MAC address (clagd-sys-mac). Run these commands on both switches in the MLAG pair.

NCLU commands

Run the net add vlan <vlan> address-virtual <anycast-mac> command. For example:

cumulus@leaf01:~$ net add vlan 4001 address-virtual 44:38:39:FF:40:94
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
Linux commands

Edit the /etc/network/interfaces file and add address-virtual <anycast-mac> under the SVI. For example:

cumulus@leaf01:~$ sudo nano /etc/network/interfaces
...
auto vlan4001
iface vlan4001
    address-virtual 44:38:39:FF:40:94
    vlan-id 4001
    vlan-raw-device bridge
    vrf turtle
...

In Cumulus Linux 3.7 and earlier, the hwaddress command is used instead of the address-virtual command. If you upgrade from Cumulus Linux 3.7 to 4.0 and have a previous symmetric routing with VXLAN active-active configuration, you must change hwaddress to address-virtual. Either run the NCLU address-virtual <anycast-mac> command or edit the /etc/network/interfaces file.

Optional Configuration

If you do not want Cumulus Linux to derive the system IP address automatically, you can provide the system IP address and system MAC address under each BGP VRF instance.

The system MAC address must be the layer 3 SVI MAC address (not the clad-sys-mac).

The following example commands add the system IP address 10.0.0.11 and the system MAC address 44:38:39:ff:00:00:

NCLU commands
cumulus@switch:~$ net add vlan 4001 hwaddress 44:38:39:ff:00:00
cumulus@switch:~$ net add bgp vrf vrf1 l2vpn evpn advertise-pip ip 10.0.0.11 mac 44:38:39:ff:00:00  
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
vtysh commands
cumulus@switch:~$ sudo vtysh

switch# configure terminal
switch(config)# router bgp 65000 vrf vrf1
switch(config)# address-family l2vpn evpn
switch(config)# advertise-pip ip 10.0.0.11 mac 44:38:39:ff:00:00
switch(config-router-af)# end
switch# write memory
switch# exit
cumulus@switch:~$

The system IP address and system MAC address you provide take precedence over the addresses that Cumulus Linux derives automatically.

Disable Advertise Primary IP Address

Each switch in the MLAG pair advertises type-5 routes with its own system IP, which creates an additional next hop at the remote VTEPs. In a large multi-tenancy EVPN deployment, where additional resources are a concern, you might prefer to disable this feature.

To disable Advertise Primary IP Address under each tenant VRF BGP instance:

NCLU commands
cumulus@leaf01:~$ net del bgp vrf vrf1 l2vpn evpn advertise-pip
cumulus@leaf01:~$ net pending
cumulus@leaf01:~$ net commit
vtysh commands
cumulus@switch:~$ sudo vtysh
switch# configure terminal
switch(config)# router bgp 65000 vrf vrf1
switch(config)# address-family l2vpn evpn
switch(config)# no advertise-pip
switch(config-router-af)# end
switch# write memory
switch# exit
cumulus@switch:~$

Show Advertise Primary IP Address Information

To show Advertise Primary IP Address parameters, run the NCLU net show bgp l2vpn evpn vni <vni> command or the vtysh show bgp l2vpn evpn vni <vni> command. For example:

cumulus@switch:~$ sudo vtysh
switch# show bgp l2vpn evpn vni 4001
VNI: 4001 (known to the kernel)
 Type: L3
 Tenant VRF: vrf1
 RD: 10.0.0.11:2
 Originator IP: 10.0.0.112 🡨 Anycast IP
 Advertise-gw-macip : n/a
 Advertise-pip: Yes
 System-IP: 10.0.0.11
 System-MAC: 44:38:39:ff:00:00
 Router-MAC: 44:01:02:ff:ff:01
 Import Route Target:
  5586:4002
 Export Route Target:
  5586:4002
switch#

To show EVPN routes with Primary IP Advertisement, run the NCLU net show bgp l2vpn evpn route command or the vtysh show bgp l2vpn evpn route command. For example:

cumulus@switch:~$ sudo vtysh
switch# show bgp l2vpn evpn route
 ...
Route Distinguisher: 10.0.0.11:2
*> [5]:[0]:[24]:[81.6.1.0]
                    10.0.0.11                0             0 5541 i
                    ET:8 RT:5586:4002 Rmac:44:38:39:ff:00:00
 ...
Route Distinguisher: 10.0.0.11:3
*> [2]:[0]:[48]:[00:02:00:00:00:2e]:[32]:[45.0.4.2]
                    10.0.0.11                          32768 i
                    ET:8 RT:5586:1004 RT:5546:4002 Rmac:44:38:39:ff:00:00

To show the learned route from an external router injected as a type-5 route, run the NCLU net show bgp vrf <vrf> ipv4 unicast command or the vtysh show bgp vrf <vrf> ipv4 unicast command. For example:

cumulus@switch:~$ net show bgp vrf <vrf> ipv4 unicast
...
 Network          Next Hop            Metric LocPrf Weight Path
*> 10.0.0.0/8    10.0.0.42                0             0 5541 I

Caveats

VXLAN Decapsulation on Maverick and Broadcom Trident II Switches

On the Broadcom Trident II+ and Maverick-based switch, when a lookup is done after VXLAN decapsulation on the external-facing switch (the exit or border leaf), the switch does not rewrite the MAC addresses or TTL. For through traffic, packets are dropped by the next hop instead of correctly routing from a VXLAN overlay network into a non-VXLAN external network (such as the Internet). This applies to all forms of VXLAN routing (centralized, asymmetric, and symmetric) and affects all traffic from VXLAN overlay hosts that need to be routed after VXLAN decapsulation on an exit or border leaf. This includes traffic destined to external networks (through traffic) and traffic destined to the exit leaf SVI address. To work around this issue, modify the external-facing interface for each VLAN sub-interface on the exit leaf by creating a temporary VNI and associating it with the existing VLAN ID.

Example Workaround

For example, if the expected interface configuration is:

auto swp3.2001
iface swp3.2001
    vrf vrf1
    address 10.0.0.2/24
# where swp3 is the external facing port and swp3.2001 is the VLAN sub-interface

auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge ports vx-4001
    bridge-vids 4001

auto vx-4001
iface vx-4001
    vxlan-id 4001
    <... usual vxlan config ...>
      bridge-access 4001
# where vnid 4001 represents the L3 VNI

auto vlan4001
iface vlan4001
    vlan-id 4001
    vlan-raw-device bridge
    vrf vrf1

Modify the configuration as follows:

auto swp3
iface swp3
    bridge-access 2001
# associate the port (swp3) with bridge 2001

auto bridge
iface bridge
    bridge-vlan-aware yes
    bridge ports swp3 vx-4001 vx-16000000
    bridge-vids 2001
# where vx-4001 is the existing VNI and vx-16000000 is a new temporary VNI
# this is now bridging the port (swp3), the VNI (vx-4001),
# and the new temporary VNI (vx-16000000)
# the bridge VLAN ID is now 2001

auto vlan2001
iface vlan2001
    vlan-id 2001
    vrf vrf1
    address 10.0.0.2/24
    vlan-raw-device bridge
# create a VLAN 2001 with the associated VRF and IP address

auto vx-16000000
iface vx-16000000
    vxlan-id 16000000
    bridge-access 2001
    <... usual vxlan config ...>
# associate the temporary VNI (vx-16000000) with bridge 2001

 auto vx-4001
iface vx-4001
    vxlan-id 4001
    <... usual vxlan config ...>
    bridge-access 4001
# where vnid 4001 represents the L3 VNI

auto vlan4001
iface vlan4001
    vlan-id 4001
    vlan-raw-device bridge
    vrf vrf1

If you use an MLAG pair instead of a single exit/border leaf, add the same temporary VNIs on both switches of the MLAG pair.

Centralized Routing with ARP Suppression Enabled on the Gateway

In an EVPN centralized routing configuration, where the layer 2 network extends beyond VTEPs, (for example, a host with bridges), the gateway MAC address is not refreshed in the network when ARP suppression is enabled on the gateway. To work around this issue, disable ARP suppression on the centralized gateway.

Type-5 Routes and ECMP

For VXLAN type-5 routes, ECMP does not work when the VTEP is directly connected to remote VTEPs. To work around this issue, add an additional device in the VXLAN fabric between the local and remote VTEPs, so that local and remote VTEPs are not directly connected.

EVPN BUM Traffic with PIM-SM

Without EVPN and PIM-SM, HER is the default way to replicate BUM traffic to remote VTEPs, where the ingress VTEP generates as many copies as VTEPs for each overlay BUM packet. This might not be optimal in certain deployments.

The following example shows a EVPN-PIM configuration, where underlay multicast is used to distribute BUM traffic. A multicast distribution tree (MDT) optimizes the flow of overlay BUM in the underlay network.

In the above example, host01 sends an ARP request to resolve host03. leaf01 (in addition to flooding the packet to host02) sends an encapsulated packet over the underlay network, which is forwarded using the MDT to leaf02 and leaf03.

For PIM-SM, type-3 routes do not result in any forwarding entries. Cumulus Linux does not advertise type-3 routes for a layer 2 VNI when BUM mode for that VNI is PIM-SM.

EVPN-PIM is supported on Broadcom Trident3 Trident 2+ switches.

Configure Multicast VXLAN Tunnels

To configure multicast VXLAN tunnels, you need to configure PIM-SM in the underlay:

The configuration steps needed to configure PIM-SM in the underlay are provided in Protocol Independent Multicast - PIM.

In addition to the PIM-SM configuration, you need to run the following commands on each VTEP to provide the VNI to MDT mapping.

NCLU Commands

Run the net add vxlan <interface> vxlan mcastgrp <ip-address> command. For example:

cumulus@switch:~$ net add vxlan vxlan1000111 vxlan mcastgrp 239.1.1.111
Linux Commands

Edit the /etc/network/interfaces file and add vxlan-mcastgrp <ip-address> to the interface stanza. For example:

cumulus@switch:~$ sudo vi /etc/network/interfaces
...
auto vxlan1000111
iface vxlan1000111
  vxlan-id 1000111
  vxlan-local-tunnelip 10.0.0.28
  vxlan-mcastgrp 239.1.1.111

Run the ifreload -a command to load the new configuration:

cumulus@switch:~$ ifreload -a

One multicast group per VNI is optimal configuration for underlay bandwidth utilization. However, you can specify the same multicast group for more than one VNI.

Example Configuration

The following example shows an EVPN-PIM configuration on the VTEP, where:

VTEP /etc/frr/frr.conf file
cumulus@switch:~$ sudo cat /etc/frr/frr.conf
...
ip pim rp 192.168.0.1
ip pim keep-alive-timer 3600
...
vrf vrf1
 vni 104001
 exit-vrf
!
vrf vrf2
 vni 104002
 exit-vrf
!
interface swp1
 description swp1 -&gt; leaf-11&#39;s swp3
 ip ospf network point-to-point
 ip pim
!
interface swp2
 description swp2 -&gt; leaf-12&#39;s swp3
 ip ospf network point-to-point
 ip pim
!
interface swp3
 description swp3 -&gt; host-111&#39;s swp1
!
interface swp6
 description swp6 -&gt; host-112&#39;s swp1
!
#auto-generated interface
interface ipmr-lo
 ip pim
!
interface lo
 ip igmp
 ip pim
!
router bgp 650000
 bgp router-id 10.0.0.28
 bgp bestpath as-path multipath-relax
 bgp bestpath compare-routerid
 neighbor RR peer-group
 neighbor RR remote-as internal
 neighbor RR advertisement-interval 0
 neighbor RR timers 3 10
 neighbor RR timers connect 5
 neighbor 10.0.0.26 peer-group RR
 neighbor 10.0.0.26 update-source lo
 neighbor 10.0.0.27 peer-group RR
 neighbor 10.0.0.27 update-source lo
 !
 address-family ipv4 unicast
  redistribute connected
  maximum-paths ibgp 16
 exit-address-family
 !
 address-family l2vpn evpn
  neighbor RR activate
  advertise-all-vni
 exit-address-family
!
router ospf
 ospf router-id 10.0.0.28
 network 10.0.0.28/32 area 0.0.0.0
!
line vty
 exec-timeout 0 0
!
end
VTEP /etc/network/interfaces file
cumulus@switch:~$ sudo cat /etc/network/interfaces
auto lo
iface lo
    address 10.0.0.28/32
# The primary network interface
auto eth0
iface eth0 inet dhcp

auto swp1
iface swp1
    link-speed 10000
    link-duplex full
    link-autoneg off
    address 10.0.0.28/32

auto swp2
iface swp2
    link-speed 10000
    link-duplex full
    link-autoneg off
    address 10.0.0.28/32

auto swp3
iface swp3
    link-speed 10000
    link-