To perform the steps, Hitless failover (of traffic, not device) is performed on one of the MCT nodes (which requires code upgrade). Traffic is completely switched over to the other node. However, the CCP will go down and come back up again once the hitless failover is completed.
MCT (multi-chassis trunking) does not support hitless upgrades of devices within the MCT topology. However, it is possible to avoid interruptions of traffic flow when upgrading MCT devices. To do this, you must first issue theclient-shutdownon the device that is being upgraded. This forces all traffic to the other MCT devices. Once the traffic is redirected, perform the upgrade using the standard upgrade procedure, and reload the MCT device while it is still in shutdown mode. When the upgrade is complete, remove the client-shutdown by entering theno client-shutdowncommand and resume forwarding traffic. The commands for this process are shown here:
MLXe1(config)# cluster MCT 1
MLXe1(config-cluster-MCT)# client-interfaces shutdown
Perform the code upgrade at this point, as there is no traffic flowing through this device.When hitless failover happens on a Brocade MLXe, NetIron MLX, or NetIron XMR node, that node flushes all the MACs and will reestablish cluster CCP session. In this case, the user may notice some traffic impact.
Once the upgrade is successfully done on this node, run the command:
MLXe1(config-cluster-MCT)# no client-interfaces shutdown
Thiswill bring up all the connections to MLXe1 and it will start forwarding the traffic. After this is completed, execute the command in a similar way on MLXe2.