This is a high level description of a problem.
We usually run a two node cluster.
This week we had an issue where one node needed to be taken down. It became non responsive and upon reboot the other node no longer functioned correctly.
So one node was left running until the maintenance window.
Apparently when it's brought back up it has the MAC of the second node in the arp cache.
This leads to node1 not being able to join the cluster.
I've seen workarounds that involve refreshing the arp cache but is there anything else to this?