SQL Server: Multiple Availability Groups, one cluster, how do I keep the primary cluster node as primary?

Question

I have a database server setup where there are three SQL Server database instances set up on three nodes of a Windows Server Failover Cluster (WSFC).

The sysadmin, who is responsible for the Windows Cluster, is accusing me of manually setting the primary cluster node as something other than what it is.

Can failing Availability Groups over to secondary replicas, to include during reboots when they automatically take over, also cause the underlying Windows Server Failover Cluster to switch the primary node and not switch it back?

This has the potential to become a distracting issue, keeping me from getting work done. I'm hoping to be pointed in the right direction to keep this from happening again.

Answer 1

Converted from comments:

Are you referring to the core cluster resources? If yes, the answer is no, failover of the cluster role for the availability group would not impact the core cluster resources. The core cluster resources will of course fail over to another node if the current owner is rebooted or loses connectivity, but that behavior is not impacted by what anything that happens with the AG resources. – Tony Hinkle

Answer 2

In order to have the failover cluster role fall-back to the "primary" node, you need to:

1. Configure the primary node. Right-click the cluster role. Click "Properties". On the "General" tab, configure the "Preferred Owners" such that the desired primary node is the only "preferred" owner.

2. Switch to the "Failover" tab, and choose the option for "Allow Fallback", and set the desired options about when to fallback.

Answer 3

Compromise achieved, after a wasted morning. King Sysadmin decided that nothing could ever be on C as a primary, not even for a moment. Even my compromise of automatic jobs that would email all of us was refused.

Having said that, I set C as manual failover, so now when A reboots, all AGs (currently five of them) fail over to B. He's okay with me manually failing back over to A when making sure that the databases are okay.