General questions

  • We are preparing to upgrade a production cluster server to SP4.  Since the only way to remove SP4 is to uninstall/reinstall SQL Server, and since only one of the DBAs here has experience with clusters, we had a test box set up to practice on before touching production.  I am starting this thread so I can post questions as they arise.  For instance, I remove the instance from one virtual server, following the BOL 'How to remove a failover clustered instance (Setup)' instructions.  This remove the instance, but left the virtual server installed.  Now, when I run setup to re-install the instance back on the virtual server, I select 'virtual server' as the server i wish to manage, and it tells me that a virtual server of that name already exists on the network.  If we need to back out SP4, and have to uninstall SQL Server on the cluster, do we have to manually delete the virtual server components?

  • Not quite sure. The way I handle SPs for SQL 2000 is I take db backups of all dbs including master.model.msdb. THEN... I stop SQL Server and all of the services in CLuster admin, goto the data folder and make a copy of the .mdf and .ldf of master/model/msdb (the app dbs too if they are small) and save them off somewhere. That way, if for some reason you have to recover everything when you reinstall instead of doing the rebuild master/recover master/model/msdb you can simply install SQL Server, the service pack you were running and then stop SQL Server, copy over the .mdf and .ldf files and you are back up and running. If you did not make copies of the app. dbs then they will be in suspect mode and you will have to recover them. But, it is simpler this way than recovering master I think.

  • We are planning on coping all database file before we do the SP install, however, we don't know quite how to rebuild the cluster short of wipinh the box and starting from scratch.  That is the reason for my question.  how do I uninstall, then re-install SQL Server on a cluster?

  • "and it tells me that a virtual server of that name already exists on the network."

    Oh boy, do I know the answer to that one.

    The Cluster installer wants to register the node's IP address and hostname with the DNS.  You need remove the hostname (A records) from DNS and try again.

     

     

     

  • So, it does not have anything to do with the IP address and network name still being in the Cluster Administrator?

  • hmmm.  I'm not sure. I just remember the cause of the error message.  I'm not sure what the installer is going to do if some of the resources already exist in the cluster group.

    I suppose you could try checking the DNS and seeing if the A records are still there. 

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

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