Hello,
I'm seeing some behavior with AlwaysOn that I'm not sure is intended with SQL Server 2012 running on Windows 2012 R2.
I have two nodes in an AG (synchronous commit with automatic failover) that are in different subnets, so the cluster and listener objects have two IPs. I'm using a file share witness for quorum. When the file share path becomes unavailable for a short period of time on either member server, the cluster automatically fails over its core resources to the host that was originally offline, but the AlwaysOn AG doesn't automatically failover.
This leads to the cluster object being active in one subnet and the listener being active in the other. Does this inherently cause any issues?
I've noticed that if I do a manual failover on the AG through SSMS, it makes the cluster and listener objects active in the same subnet so I assumed that's how it 'should be'. I honestly haven't noticed any issues with data availability when it gets in the state with the cluster and listener objects active in opposite subnets, but SCOM alerts that the cluster is offline and I do notice DCOM errors in the event logs of the primary replica when it gets in that state.
Any input is appreciated.