I'm hoping someone can help bring some sanity to what I'm experiencing.
I have 2 SQL Server 2016 instances clustered with windows clustering. There are 11 user databases on the instance, all set up in their own availability group. All 11 availability groups are set up the same. They are set up for asynchronous
commit, manual failover, non-readable secondary. There is *not* an availability group listener set up for any of the groups. Basically, we purposefully set these up to need manual intervention for anything. All of the applications have connection
strings pointing to the primary node ONLY.
Last weekend, we had some maintenance on our primary node and it was bounced. During that time, there are a flurry of login attempts by all applications to our secondary node. This was surprising, but we assume the cluster was routing connections
during the downtime. After the primary node came back online, most of the applications stopped trying to connect to the secondary node.
However, we're still seeing one application that will attempt to connect to the secondary node several times per day. It's very random...there will be 3 attempts, then a few hours later 1 attempt, then a few hours later 10 attempts, etc. This behavior
has persisted through an application reboot. The application logs are not showing any failed login errors or really any errors that correlate to the attempted login time.
The only theory we have is that SQL Server (or the windows cluster or the application itself querying) is making the availability group known to the application and for some reason the application is trying to connect there.
I've reached out to the application vendor who claims that the application will only try to connect to the host configured in the connect string....
Anyway, if anyone out there could give me something to try I would be very grateful.