locked
Different Node State+Health reported to different administrators (known bug?) RRS feed

Answers

All replies

  • Hello Christian,

    You can use the following:

    cluscfg view /scheduler:<clustername>

    or

    powershell cmdlet
    get-hpcnode

    On the console of the HPC Head node, does it show the compute nodes unreachable?

    Thanks,
    Ben
    Wednesday, July 15, 2009 7:23 PM
  • Well, we fixed this issue as described here: http://social.microsoft.com/Forums/en-US/windowshpcitpros/thread/a9075106-b474-4ca4-9dd8-02bcf529211c (MWirtz is my colleague), but it is not "nice" to alter the database manually. We ran into this issue a couple of times already, that is why I asked whether there is a known problem. A similar (maybe the same) issue has been discussed here: http://social.microsoft.com/Forums/en-US/windowshpcitpros/thread/916ec675-e9ff-4ed9-82aa-065a4b766312.

    We just applied SP1, maybe it helps.

    Regarding your question: The HPC Cluster Manager running on the Head Node seems to always display the correct node state (same for the command line interface). But getting wrong results when running an administrative tool on the client is certainly not of much value. I cannot state whether the comand line tools report the right results when used on a node other than the head node.

    Our workaround is to just use the HPC Cluster Manager on the Head Node.


    Kind regards,
    Christian
    • Marked as answer by Don Pattee Saturday, July 25, 2009 6:29 AM
    Monday, July 20, 2009 7:48 AM
  • Christian,

    Thanks for the reply. I am curious if you are still getting Access Denies within your logs from time to time which may lead up to problems in the db. Are you still seeing errors relate to the following?

    Access is denied to user 'NT AUTHORITY\ANONYMOUS LOGON'.


    Thanks,
    Ben
    Tuesday, July 21, 2009 5:41 PM