January 26, 2015 at 1:52 pm
We have a number of databases on an SQL2008R2 server which are mirrored. There is no witness configured and the mirroring is setup for High Safety without automatic failover.
The mirror is currently in place for data redundancy and no application connection strings have been explicity configured with the failover partner name, however my understanding is that the SQL Native Client will pick up the mirror partner configuration and cache it in case a failover occurs.
Let's refer to ServerA as the pincipal and ServerB as the mirror.
Our issue is that around the same time each night (coinciding with backups and therefore resource strain etc) we are seeing the applications which are normally using ServerA suddenly attempting to connect to ServerB which fails as the database on ServerB has not failed over and is in a Restoring... state. Obviously this manifests as errors and timeouts (24x7 application too). Sometimes (but not always) there will be a mirroring timeout on ServerB as it cannot ping ServerA to confirm it is up. There is no failover from ServerA to ServerB
There is nothing obvious in the server logs on ServerA that shows any issues with the server itself. We suspect that ServerA is becoming unresponsive due to load at that time. The mirroring timeout is currently set to 90.
Clearly there are issues with access to ServerA from the client and on occassion the mirror, but my questions are:
- Does the SQL Native Client indded cache the failover partner after a successful connection to ServerA and then attempt to connect to ServerB, even though the failover partner is not initially set in the connection string?
- What scenarios will cause the client to switch over to ServerB?
- What tools/approaches could we use to confirm exactly what is happening here?
- Any ideas for workarounds?
Thanks in advance...
Viewing 0 posts
You must be logged in to reply to this topic. Login to reply