none
How to clean the cluster model?

    Question

  • Recently one of our compute nodes died. Now it shows as unreachable in the compute cluster management console, and when I try to remove it from the cluster, it does not appear in the list of currently available nodes.

     

    How would one generaly clean the cluster model and remove dead records from it ?

     

    Regards

    Ivan

    • Moved by Josh Barnard Thursday, March 26, 2009 12:38 AM (Moved from Windows HPC Server Developers - General to Windows HPC Server Deployment, Management, and Administration)
    Wednesday, February 06, 2008 11:54 AM

Answers

  •  

    Hi Ivan,

     

    It appears your Scheduler DB and SDM are out of sync.  This issue is a bit more complicated than can be handled by a newsgroup, so I'd have to refer you to product support.

     

    If it is any consolation, this behavior has been handled much better in our newest product, Windows HPC Server 2008.

     

    Information on contacting support is at https://support.microsoft.com/oas/default.aspx?c1=508&gprid=11782.

     

    Best Regards,

    Jarred Clore

    • Marked as answer by Josh Barnard Thursday, March 26, 2009 12:38 AM
    Wednesday, April 09, 2008 8:00 PM

All replies

  •  

    Hi Ivan,

     

    Is this a Compute Cluster Server (v1) or HPC Server 2008 (v2, in beta) product?

     

    Thanks,

    Jarred Clore

    Tuesday, February 12, 2008 9:46 PM
  •  

    Jarred,

     

    It is a version 1 of the compute cluster pack.

     

    Regards

    Ivan

    Monday, March 17, 2008 5:06 PM
  • Hi Ivan,

     

    I'm not sure I'm clear on exactly what you wanted to do . . . are you trying to permanently remove the node?  Or bring it back into the cluster?

     

    To remove it, highlight it in the Administrator Console and you should see a "Remove" command in the Action Pane.  Is that not working for you?

     

    To add back in . . . just get it up and running again, install the Cluster Pack and select the Compute Node role during installation . . . it should contact the HN and rejoin when it comes online.

     

    Thanks,
    Josh

    Friday, March 28, 2008 10:33 PM
  • Josh,

     

    The node physically does not exist. When I use the Remove command, I get a wizard which lists all the nodes in the cluster, but not this one. So I cannot choose to Remove it.

     

    Regards

    Ivan

     

    Monday, March 31, 2008 7:45 AM
  •  

    Hi Ivan,

     

    It appears your Scheduler DB and SDM are out of sync.  This issue is a bit more complicated than can be handled by a newsgroup, so I'd have to refer you to product support.

     

    If it is any consolation, this behavior has been handled much better in our newest product, Windows HPC Server 2008.

     

    Information on contacting support is at https://support.microsoft.com/oas/default.aspx?c1=508&gprid=11782.

     

    Best Regards,

    Jarred Clore

    • Marked as answer by Josh Barnard Thursday, March 26, 2009 12:38 AM
    Wednesday, April 09, 2008 8:00 PM