Hi, all.
I want to bring this up front before I begin. I am, by no means, an expert at clustering, or SQL Server. However, I want to try and figure out why it takes so long for our tools to connect to the database server after a fail over.
Background:
OS: Server 2008
SQL Server 2005
2 node cluster server w/ NETAPP SAN (connected to each node using iScsi connection)
Observations:
- Happens whether fail-over is REAL or simulated (simulated being through the cluster manager).
- When fail over occurs, all services go offline.. and then come back online in less than a minute (~45-60 seconds).
- Active node does switch successfully when fail over-occurs.
- I can see the cluster storage / SAN (storage drive, Qorom, etc.) on the active node.
- I am able to access the database, local.. on either node, using Enterprise Management Studio.
- What I can't do is access the database from another computer outside the network for a certain period of time
- Time that the database is unavailable can range from 6 minutes to up to 13 minutes. However, I think the time differences here are simulated fail-over events (cluster manager) vs. forced fail-over events (shutting down active node).
- using NETSTAT -A, I could not see anything as far as lost packets or collisions.
- Ping test is fine from server to clients.
- All clients outside of the subnet, but within the factory network.
- When the clients are connected to the server's database, there are no issues with connectivity.
- This "pause" in DB connection from outside clients was not happening when it was 1st installed roughly a year ago (So, I am GUESSING it's not a setup or configuration issue).
I can't just assume this is a network issue because my equipment is always "guilty until proven innocent." At least in the IT department's eyes.. :), so I want to exhaust all possibilities that this could be an issue with the server itself.
Can anyone think of anymore tests to try to rule out the server? I'm all ears. Please remember, I am not an expert, so sorry if I ask how to do half of the actions you suggest... :)
Thank you so much for any suggestions!
Best Regards,
Bill