Failback will not work

  • We have Win2K Advanced, SQL 2K Enterprise setup in a cluster and failover would work flawless between Node A and B without a problem. Did testing with it for about 3

    hours without any problems. After installing SP1 for SQL2k and rebooted both servers, we can failover to Node B, but can't fail back to Node A. When we try to failback to Node A the cluster goes offline. No matter how we try it (manual move or restarting Node B) the cluster goes

    offline. I've uninstalled the entire solution, reinstalled back to the same point and have the same results as before -- everything works fine until SP1 gets

    installed. Am I missing something here?!?!When you go into the event viewer you will see that the resource is failing, loses network connectivity with the other Node

    and then re-establishes the connection. Followed all the documentation on this, and yet it won't work. I would like to go with Active/Active, but my client does not have the funds to do this, so Active/Passive is what we are stuck with.

    Please help if you can -- I can't seem to figure this out.


    Thats Bigman 2 Y-O-U!

  • I have a gut feeling that one node didn't get upgraded. Look at this technet article and reapply the SP1

    http://support.microsoft.com/support/kb/articles/Q306/0/34.ASP

    Wes

  • Thanks Wes. Unfortunately the technet article and the workaround that they have listed is not working for us. I'm not finding the file that they are speaking of so I guess I am lost again. I am going to try and uninstall the SQL Cluster and see if the Windows Cluster works now -- once I have verified this, I will try to reinstall SQL back into the cluster and see what happens. I will keep the forum posted and let you know. Thanks again.


    Thats Bigman 2 Y-O-U!

  • Well tried to remove and add the SQL Server cluster back in and we have the same result. I just don't understand what could be causing the problem. Everything works 100% and when Service Pack 1 is installed -- BOOM....it will not work any more. Any ideas?


    Thats Bigman 2 Y-O-U!

  • Ok, now I don't know if this makes a difference, but speaking with another engineer he says he shouldn't. I followed Microsoft's advice for setting up a domainlet and everything is working. However, going through my Event Logs for Directory Service whenever I try to failover to the other node, I receive this error message:

    Unable to establish connection with Global Catalog.

    Now according to the domainlet installation, you shouldn't have to have the Global Catalog. Is this correct or not? Basically no users will be authenticating in the domain -- it will just be for SQL. Do you still have to implement the Global Catalog?


    Thats Bigman 2 Y-O-U!

  • I don't understand what domainlet is. When you do the cluster install I would make sure both machines are apart of the domain. Secondly, Make sure all the services start as a domain account. Have you looked at the cluster installation guide posted here?

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply