6120 link down on mgmt0 doesn't trigger UCS Manager cluster failover - should it?

[cnicho25]

Hello Friends

We have recently installed a UCS cluster consisting of two 6120s configured for HA. While performing fail over testing scenarios, we removed the network cable in mgmt0 on the primary 6120. Immediately the cluster IP address stopped responding to ping (as expected) and we lost connectivity to the UCS Manager GUI (also as expected.) At some point however we were expecting the subordinate 6120 to detect this link was down and to initiate a fail over of UCS. This did not occur after 20 minutes of waiting.

My questions are as follows:

1. Is this expected behaviour?
2. If not, what should we be looking at to make sure fail over does occur in the future?

I know we can force a failover from the subordinate by issuing a cluster lead command from local-mgmt but would be interested to see if it should be occuring automatically on link failure of primary mgmt0.

Thank you for your time.

[Robert Burns · Accepted solution]

Configured as default that is expected behavior.

You can configure management interface monitoring to switchover if the management interface loses connectivity such as your test case.  This is what you're looking for.

Admin Tab - Communication Management - Management Interfaces - Management Interface Monitoring Policy Tab.

Regards,

Robert

[cnicho25]

Fantastic thanks so much for the fast reply Robert.