0

We have created an Always on SQL Failover Cluster. There are 4 servers in the cluster, 2 on one subnet and 2 on another.

FAILOVER CLUSTER (172.16.82.22 OR 172.16.92.22) SQL AVAILABILITY GROUP LISTENER (172.16.82.23 OR 172.16.92.23)

When we initiate a SQL Failover from subnet A to B, the listener resource automatically rolls over to the 92 subnet. But the cluster availability group stays on the 82 subnet which then causes things to not connect as SQL is waiting for the other resource to join the 92 subnet. All the servers are in the preferred owner groups.

I should also mention that i can manually get the availability group to failover by going to More Actions > Move Core Cluster Resources > Select Node and then picking a server on the site B subnet.

What could be my issue here?

David Hanson
  • 1
  • 2
  • To be clear, it sounds like you have an Availability Group (given the mention of an AG listener), not a Failover Cluster Instance (you've got a mix of terminology in your question). How are you performing the AG failover? That is, are you doing `alter availability group [foobar] failover;` or are you initiating it via cluster manager? Also, something seems weird about how you're describing things - the AG itself and its listener should be in the same cluster role and as such should not be able to be owned by different nodes. – Ben Thul Aug 12 '22 at 17:22
  • Thank you for the response. Sorry if i was being confusing while i described things. We have the Windows Failover Cluster, the cluster has 2 IP addresses defined (82.22 & 92.22). In roles we have the SQL Availability Group, which includes the listener. (82.23 & 92.23). In nodes we have the 4 SQL server members. When we failover the listener from the 82 to the 92 subnet, we need the windows cluster to failover to the matching subnet at the same time. – David Hanson Aug 14 '22 at 16:53
  • No problem. The role that contains the cluster IP and the role that contains the AG listener are independent. That is, it's is a perfectly valid config to have the cluster role on Node A and the AG role on Node C (both of which can be on different subnets). If you want to, you can fail over the cluster role to match the home of the AG role. But the only time I've ever really worried about where the cluster role itself lives is when I'm doing firmware patching on a particular node. That is, in day to day operations, I leave the cluster role alone. – Ben Thul Aug 15 '22 at 03:18

0 Answers0