SQL Server Performance Forum – Threads Archive

Cluster Admin and command don’t work

I have a 2 node SQL cluster in Active Passive configuration. When we run SQL cluster from 2nd node, we get strange behavior and that is we get intermittent errors: SqlDumpExceptionHandler: Process 57 generated fatal exception c0000005 EXCEPTION_ACCESS_VIOLATION. SQL Server is terminating this process. Though everything works fine. But now we are observing additional problems and that is when above error happens, the cluster administration will not work or cluster commands will not provide information about the cluster. And, one can’t telent to the box or use terminal services and in addition net view from another box will not show 2nd node. The only option to recover is reboot. The SQL is running on Win2k Adv with Win2k SP3 and SQL2000 with SP2, AWE and 8GB. Any insight will be high appreciated and thanks in advance! Regards

Is this a new installation? Has node 2 every worked correctly? Is all the hardware on Microsoft’s HCL. Do you have all the latest hardware drivers? I assume SQL Server 2000 is the Enterprise Version. I assume you followed the steps described on this website to install your cluster. I know these are all basic questions, but in my experience, it is almost always a very basic issue that causes strange problems. I don’t have enough data to provide more help at this time. My guess is that you will need to verify everything I described above, and then reinstall both Windows and SQL Server clustering to resolve the problem. Also, based on my experience, a reinstall is the only way to solve most prolems, as hacks usually don’t work. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

This was a brand new installation but has been running for more than 4 months. The SQL2000 EE is used and hardware is on MS’HCL. The behavior has been noticed on 2nd node only. The SQL VS works fine on Node1. Two months ago cluster was failed over to 2nd node and after running for 10business days, we received the first access_violation error. But now not only we get error message frequently if we run SQL VS on 2nd node but also remote connectivity fails too.

I have checked out this error message on the web, and in general, this kind of error results from a bug (but not always). First, you may want to add SQL Server SP3 to your cluster, and if that doesn’t fix the problem, you may want to open a support issue with Microsoft PSS. If it is a new bug, then PSS will help you resolve it. Also, be sure that the hardware, drivers, software, and configuration settings are identical on both servers. If they are, then most likely, and you have upgraded to SP3, then most likely you have discovered a bug. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

An AV indicated a bug. Your best bet is to contact MS PSS.
_________
Satya SKJ
Moderator
SQL-Server-Performance.Com

Thank you and I feel the same. I have opened a case with MS and will post the resolution. Meantime, any additional information… is always welcome. Regards

hi vinny, i’m experiencing the same problem. did you get any resolution from microsoft? Raymond

Are your servers in the same domain or just in a workgroup.

quote:Originally posted by Vinny I have a 2 node SQL cluster in Active Passive configuration. When we run SQL cluster from 2nd node, we get strange behavior and that is we get intermittent errors: SqlDumpExceptionHandler: Process 57 generated fatal exception c0000005 EXCEPTION_ACCESS_VIOLATION. SQL Server is terminating this process. Though everything works fine. But now we are observing additional problems and that is when above error happens, the cluster administration will not work or cluster commands will not provide information about the cluster. And, one can’t telent to the box or use terminal services and in addition net view from another box will not show 2nd node. The only option to recover is reboot. The SQL is running on Win2k Adv with Win2k SP3 and SQL2000 with SP2, AWE and 8GB. Any insight will be high appreciated and thanks in advance! Regards

1. Please reapply SQL service pack again when both nodes are available in the cluster. 2. Reapply all SQL hotfixes with both nodes up and running in the cluster. 3. For non cluster aware hotfixes: If hotfixes are not cluster aware then reapply them with SQL VS running individually from each cluster node.

]]>