When configuring MLAG, pay attention to the following notes:
- When peer-link is down, but both the MLAG member ports are up, the split-brain failure scenario occurs. The MLAG system will operate abnormally and cannot be automatically recovered in this scenario. To ensure the reliability of peer-link, note the following points when configuring and deploying the peer-link:
- When configuring the peer link, only the LAG port is supported. In other words, the peer link must always be a LAG port.
- A minimum of two directly connected physical ports should be used for peer link LAG port. The peer link connecting the two MLAG peer devices should be directly connected meaning that no intermediate transmission devices should be used on the peer link. All of the directly connected physical ports should be added into one LAG port to form the peer-link. We don’t support more than one L2 connection between MLAG peer switches meaning that only one peer link is supported between the two peer devices.
- 10G or 40G speed ports should be used on peer link. It is highly recommended to provide enough bandwidth for the peer link when the network is deployed.
- Any manual action to shutdown the peer link is strictly forbidden.
- Any MLAG VLAN and non-MLAG VLAN MUST be allowed on MLAG peer-link.
- When numerous rapid PVST+ instances are configured, exceeding the default BPDU queue processing rate in CPU will result in BPDU packet loss or network loops. To resolve this problem, you can use CoPP command set class-of-service scheduler bpdu-scheduler max-bandwidth-pps <value> to increase the maximum bandwidth of BPDU queue. By default, the value is 80pps.
- When numerous MLAG instances are configured, exceeding the default MLAG queue processing rate in CPU will result in MLAG packet loss. To resolve this problem, you can use the following CoPP command to increase the maximum bandwidth for MLAG and MLAG MAC SYNC queues. By default, the value is 80pps.
set class-of-service scheduler mlag-scheduler max-bandwidth-pps <value>
set class-of-service scheduler mlag-mac-sync-scheduler max-bandwidth-pps <value>
- Please proceed with caution when removing the peer-link interface from a VLAN, it may cause problems in spanning tree calculation if proper attention is not paid to its potential ramifications.
- MLAG supports to deploy with IGMP snooping.
- MLAG supports to deploy with DHCP snooping.
- MLAG supports to deploy with rapid PVST+.
- MLAG supports to deploy with VXLAN.
- If an L3 device is dual-home attached to the MLAG domain as a leaf node, ECMP should be configured on the uplink ports of the leaf node instead of LAG.
- When deploying VRRP in MLAG topology, pay attention to following points:
- A pair of MLAG ports should be added into one VRRP instance. Multiple pairs within the same VLAN can be added into one VRRP instance as well. However, MLAG ports with different VLANs should be added into different VRRP instances.
- Active-active VRRP should be used in order to support load-balancing between the two links connected to the Access Layer device.
- After the configuration is changed from inconsistent to consistent, you need to restart the MLAG peer devices to ensure that MLAG functions normally.