I recently inherited a SQL Cluster (2008R2) which for the most part behaves itself impeccably.
The windows cluster is made up of two nodes running Active/Passive, Node1 and Node2 are dedicated blades in two different data centers. There are 3 SQL instances all running on Node1. Quorum is established by a File Share Witness and we have a heartbeat between the two nodes.
The other day someone switched off the file share witness by mistake, and the windows server failed over from Node1 to Node2. Or should I say, in Failover Cluster Manager, Node2 was now specified as the active node by Windows.
However, the SQL Cluster didn't do anything. All the instances stayed up and hosted on Node1. I would have expected them to move Nodes, but no.
There was no adverse affect on the databases at all.
Once power was resumed to the File Share Witness I brought it online again and the Windows Cluster failed back to Node1.
Our Windows Technicians are looking into why the cluster failed over, and I'm left scratching my head with the SQL bit.
All I can think of is that the heartbeat kept the SQL instances on Node1 and losing the witness wasn't important.
I'm still learning the small details of Windows Clustering, being much more used to Log Shippping and Mirroring when it comes to HA solutions, so any insight into why the SQL Instances didn't failover would be appreciated.