January 26, 2012 at 12:13 pm
We are having connection timeout issues with one node only of a two node SQL Server 2008R2 cluster running on Windows Server 2008R2. We notice the issue when users are connecting from their PCs to a database on the cluster using an ODBC connection from MS Access, but only when a particular node is the active node. They are getting a timeout error from SQL Server.
I also get a similar error when running SSMS from my desktop and attempt some action on this SQL server instance. But again this only happens when a particular node of the cluster is active.
Both nodes should be configured identically, however this connection error persists when one of the nodes is active. I don't believe it could be any setting within SQL Server itself, since the instance is clustered, but maybe there is a different system setting on one of the nodes that is causing these connection timeout issues.
January 27, 2012 at 12:27 am
I'm not sure of your issue, do *some* client connections timeout or do *all* client connections timeout when on the troublesome node?
What does
select @@version
return on each node? is it identical?
Are there differences in the sql error log when sql starts up between the different nodes? Errors or warnings?
named or default instance?
January 27, 2012 at 6:17 am
The client connections that fail are those from an MS Access front end using ODBC and from within SSMS itself. The first connection attempt tends to fail and then succeeds on the second attempt. The two nodes of have identical versions of SQL Server 2008R2. Although the firewall settings on the two nodes should be identical, I am becoming suspicious that some port might not be open on one of the nodes.
Also it could be an issue with DNS name resolution for the IP address associated with the SQL Server instance virtual name.
I am referring this to the network team for the time being to at least rule out any firewall or name resolution issues before exploring other possible causes.
Thanks!
May 14, 2012 at 1:34 am
Hello,
did you find a solution, i'm facing almost the same problem in my environment?
I replaced one virtual cluster node with a physical machine, my ERP-Software works fine on both nodes, but the ODBC connections direct to the database don't work if the cluster runs on the new node.
Maybe related to DNS i think. If i ping the SQL-Cluster name on the new node i get a time out. All other ressources i can ping without time out (Windows Cluster name, DTC Cluster name).
Thank you,
Thomas
May 15, 2012 at 6:02 am
This might help ...
http://serverfault.com/questions/179898/sql-server-timeout-on-first-attempt
Is the instance your having problems with a named instance or the default?
Best Regards
Tom.
May 15, 2012 at 7:31 am
Hello,
thank you. My problem is bit different. Not just the first call brings the error. The whole time there are errors randomly.
My Instance is a clustered instance, so it's kind of a named instance.
Are there any connection restritions in windows to limit the connections for odbc? Maybe there are too much open connections at one time.
Thomas
May 15, 2012 at 8:06 am
You can set the number of connections in server properties, you can also disable remote connections - I assume remote connections are not disabled against that server. Failing that, it could be a random port assignment that is giving connection issues due to firewall setings (server or client) perhaps.
Regards
Tom.
July 14, 2012 at 5:45 pm
thomas.gundolf (5/15/2012)
Hello,thank you. My problem is bit different. Not just the first call brings the error. The whole time there are errors randomly.
My Instance is a clustered instance, so it's kind of a named instance.
Are there any connection restritions in windows to limit the connections for odbc? Maybe there are too much open connections at one time.
Thomas
This is exactly what I am experiencing. One node seems fine, but while apps connect to the second node, they have random timeout errors and I can't stay on the second node. Very weird. The two nodes are identical in every way and there are no warnings or errors in the event logs or cluster logs.
July 16, 2012 at 12:33 am
Hello,
i open a Microsoft Call - they couldn't find anything in logs etc. So they asked me to install network monitor (netmon) to get a deeper sight into the traffic. From that moment on everything worked fine without doing anything.
Microsoft told me that while network monitor is being installed a TCP/IP Stack gets reloaded/reinstalled/flushed or something like that - an that fixed my problem.
Hope you find a solution too...
Thomas
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply