October 27, 2007 at 2:51 pm
Hi
I'm currently facing a cluster issue and I can't think how can I fix the issue. (I didn't installed that cluster)
The cluster is in active / active mode but should be (after the fix) active / passive.
It is hosted on Win 2k3 Ent R2 Sp2.
One machine has 18 gigs of ram
the other only 2 (for the moment)
MSDTC is installed and configured adequately (like other cluster setup we've installed)
On the cluster, there's 7-8 SQL instance running at the same time.
The issue is: the first node, which has 18 gigs of ram cannot take all the instance while the one having only 2 gigs was able to (only once we don't know why we were able that time to move all the nodes).
Also when I try to move all instance on node 1 (18 gigs of ram) the two last one to switch node will always fails. If I close any running instance and start back one that failed, it will go up. So in short I can have only 5 running instance no matter which one is up and which one is down.
I've monitored the memory using a windows counter and I was having more than 12 gigs of ram left at any moment and the paging file was up to 40 gigs.
When trying to start back a failed instance I got in the event log the error 41d which doesn't say anything useful. Googling didn't help either.
The SQL log file never got anything about that failed start.
All account have AD admin privileges.
Windows was granted the /PAE option to take advantage of more ram and the policy LOCKING MEMORY was activated to enable the AWE option in SQL (which didn't help much) and so was deactivate at the end.
One thing that bugged me is the SQL version. On some machine it was the RTM version of 2005 while on the second node it was the SP2 version. So when an instance switched from node 1 on the RTM version, it was then on the SP2 version on node 2.
I don't think the mismatch version is the issue as I guess it is more cluster configuration related however it bug me if it's something that is "allowed" in sql to run node with different version.
Any help on that will be greatly appreciated.
October 27, 2007 at 8:35 pm
The fact that there is a mismatch is worrisome. It means the service pack applied correctly to one side but not the other. With these kinds of issues, your better served contacting Microsoft for support.
K. Brian Kelley
@kbriankelley
October 29, 2007 at 5:08 am
Thanks for the reply.
The service pack issue, to my knowledge, didn't installed successfully because DTC wasn't configured at all so none of the instance had a transaction coordinator available which the service pack installer need.
August 6, 2008 at 9:12 am
This was the same issue as:
http://www.sqlservercentral.com/Forums/Topic416644-149-1.aspx
Although it might seems that different version of SQL is not recommended, it did upgrade gracefully without any issue (and work correctly for those instances that were able to run outside the cluster group issue)
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply