I just upgraded my test Active/Active cluster from sql 2000 sp4a to SQL 2005. While the sql instance that normally runs on node2 was failed over to node1 instance2 refused remote connections. I have not had a chance to test failover the other way (all instances to node2). Obviously this defeats the purpose of clustering.
What am I missing in the setup or config?
Failover worked both ways prior to the upgrade.