Unable to see Cluster disk

  • We currently are running two cluster servers in production.  Since both were set up at least two-three years ago, the DBA who did the setup is uncertain of much of how the servers were setup, and the other DBAs have never done an install on a cluster.  Therefore, we have a server we are using to practice this process.

      I have installed one virtual server on the active node of the cluster with no problems, however, when I try to install a second virtual server on the same node, I cannot see the cluster disks in the install process.  At first, we only had one physical disk defined as a resource to the custer and I put the databases for the first virtual server on it.  After I was not able to do the install of the second virtual server, and thinking that the first virtual server had locked the drive, we installed a second disk and defined a drive to the clsuter.  However when I tried to install the second server again, I still do no see any cluster disks in the install app.  I referenced both BOL and Microsoft support, but neither talks about this problem.  Other than installing everything on the C: drive and then moving the files to the desired location, I do not know how to continue.  I would appreciate any suggestion.

  • Have you created the new "cluster disk" as a physical resource in the cluster administrator?

  • Yes, the new disk is visible in the Cluster Administrator as a Physical Disk resource.

  • Where is the physical disk resource located?  Did you create a resource group and then the disk resource in that group?

    I think you need to let the installer create the disk resource and resource group during the install.  I don't believe it will let you pick one that is already created in a resource group.

     

  • The disk resourceand the resource group are created by the sysadmin before the install.  We have found that if we create a second resource group and move the second disk to it, the second disk is visible.  Do we need a seperate resource group for each virtual server we wish to install on a cluster?

  • Do we need a seperate resource group for each virtual server we wish to install on a cluster?

    yes, aboslutely.

  • And it looks like I was wrong about the resource groups.  Your admin needs to create the group.  It's been a while....  SQL server installer lets you pick a group that has an available cluster resource of type physical disk.

     

     

  • Okay.  We created a second resource group and were able to install the second virtual server.  The reasons for my perplexity were, 1. the documentation does not say that you need a seperate resource group for each virtual server, and 2. the DBA and Sysadmin who did the original install went back later and moved everything into one resource group, so both virtual servers are active on the same node and everything failsover as a unit.  This setup does not seem to cause any problems.  You would think you could do this from the start.  Thanks for all your help.

  • Here's from the BOL topic "Creating a Failover Cluster"
     
    Elements of a Virtual Server

    A virtual server contains:

    • A combination of one or more disks in a Microsoft Cluster Service (MSCS) cluster group.

      Each MSCS cluster group can contain at most one virtual SQL Server.

    I have wondered about moving all of the SQL resources into a single resource group but have never actually tried it.  You might want to check with MS to see if they will support what you've done.  There may be a problem down the road.

     

     

     

     

  • Heh, missed that one line.  I was going on what I was told by another DBA and only glanced at the BOL entry on failover clustering.

  • Hi,

    if you put more then one virtual SQL Server into the same resource group you will have to failover all of them, even if only one actually fails...not something i would like to have in a production environment...

    karl

    Best regards
    karl

  • We are not as concerned with a virtual server failure, as a hardware failure.  Plus a recent path to the Distributed Transaction Co-ordinator requires that it be defined explicitly in a resource group.  Since we have links to Oracle in both our virtual servers, we had to place them into the same recource group as the DTC.

  • Sorry, I meant a recent patch to the Distributed Transaction Co-ordinator.

Viewing 13 posts - 1 through 12 (of 12 total)

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