When putting the primary node (1) into maintenance mode, the LACP fail-over performs as expected. After making the node in maintenance (1) available, connectivity is lost and does not return… lasting over 5 minutes.
To recover, I have to force node (2) into maintenance mode. Then the LACP connection is move to the correct node (1).
It appears there is a preferred master. The controller mastership is returned to the original node (1), while LACP remains connected to the secondary node (2).
Is there a setting I can change to insure that the controller mastership remains on the active node?
A50 – v18.104.22.1682
For HA or LACP. Both the Nodes in the Cluster should have identical configuration, inclusive of Network.
When the both port in LACP fails in the Primary node, Pool will be imported to Secondary Node and it will use the LACP of secondary node
When the Node 1 comes back into available, Pool will be imported back to Node1 from Node2 and now LACP will be used from Node1.
Below link provides more details of how HA works
It would be good if you can share the logs. We need to look into the logs/system to figure what happened during the HA Process. You can collect all the logs from the following path
Please send the logs to email@example.com.
I have identified and resolved the issue.
Due to the A50 technically operating as two (2) separate devices. The LACP configuration needs to have two (2) distinct port-channel groups. One (1) for each node on the physical device.
I originally build the LACP configuration using a single port-channel group. Connecting to all of the LAGG ports with the same port-channel group. Which resulted in the strange behavior mentioned above.
You must be logged in to reply to this topic.