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.
Use 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 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
/etc/network/interfaces file to add
bridge-ageing to the bridge stanza, then run the
ifreload -a command.
cumulus@switch:~$ sudo nano /etc/network/interfaces ... auto bridge iface bridge bridge-ageing 600 ...
cumulus@switch:~$ ifreload -a
Configure a Switch Virtual Interface (SVI)
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 virtual 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:
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
/etc/network/interfaces file to add the interfaces and VLAN ID you want to use, then run the
ifreload -a command. 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 ...
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.
IPv6 Link-local Address Generation
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:
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
/etc/network/interfaces file to add the line
ipv6-addrgen off to the VLAN stanza, then run the
ifreload -a command. 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 ...
cumulus@switch:~$ ifreload -a
To reenable automatic link-local address generation for a VLAN:
net del vlan <vlan> ipv6-addrgen off command. The following example command reenables 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
/etc/network/interfacesfile to remove the line
ipv6-addrgen offfrom the VLAN stanza, then run the
bridge fdb Command Output
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:
|self||The 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
|master||The 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
|extern_learn||The 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 extern_learn master bridge 02:02:00:00:00:08 dev vx-1001 dst 184.108.40.206 self extern_learn
- 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
- A bridge cannot contain multiple subinterfaces of the same port. Attempting this configuration results in an error.
- In environments where both VLAN-aware and traditional bridges are used, if a traditional bridge has a subinterface of a bond that is a normal interface in a VLAN-aware bridge, the bridge is flapped when the traditional bridge’s bond subinterface is brought down.
- You cannot enslave a VLAN raw device to a different master interface (you cannot edit the
vlan-raw-devicesetting in the
/etc/network/interfacesfile). You need to delete the VLAN and recreate it.
- Cumulus Linux supports up to 2000 VLANs. This includes the internal interfaces, bridge interfaces, logical interfaces, and so on.
- In Cumulus Linux, MAC learning is enabled by default on traditional or VLAN-aware bridge interfaces. Do not disable MAC learning unless you are using EVPN. See Ethernet Virtual Private Network - EVPN.