none
.NET installation affecting fail-over cluster RRS feed

  • Question

  • We have a recurring problem when installing .NET framework 3.0 on clustered systems. It seems the installation makes the cluster service stop, causing the cluster resources to move to the other node. This happens on a number of different systems.

    What is interesting is that the cluster manager seems to get a proper shutdown message rather than something crashing, and gracefullt shuts down the cluster services on the node. I.e., it does not seem that http://support.microsoft.com/kb/324393 is relevant, particularily as the problem remains when no IIS-related groups are running in the cluster.

    Narrowing things down, we have concluded that the problem in fact happens when XPSEPSC-x86-en-US.exe is run by the .NET fw installation. In fact, it happens during the 25 or so seconds when the following is executed by that program:

    (from the log of XPSEPSC-x86-en-US.exe \quiet \norestart)
    4.047: Starting process:  C:\WINDOWS\system32\spool\PRTPROCS\W32X86\printfilterpipelinesvc.exe /RegServer j:\cf608895074a8b6d25c32784004718b3\update
    31.813:  Return Code = 1726

    It is always during this time the cluster node stops, and it is reproducible both when executing XPSEPSC-x86-en-US.exe directly, and when installing .NET framework 3.0

    XPSEPSC is the "XML Paper Specification Shared Components Pack 1.0", which does not seem terribly related to MSCS.

    Has anybody else had similar problems? Any ideas for what might be the problem?

    Any help would be appreciated!

    /Kristian Kopsen

    ---
    The clustered systems use Windows Server 2003 R2, EE, SP2 and the clussvc.exe is of version 5.2.3790.3959
    The .Net installer is dotnetfx3.exe version 3.0.4506.30
    The XPSEPSC-x86-en-US.exe is version 6.1.22.4, included in the above
    Thursday, January 15, 2009 8:48 AM

All replies

  • After installing .NET Framework, can the cluster manager be started again?

    Thanks.
    Please remember to mark the replies as answers if they help and unmark them if they provide no help. http://forums.msdn.microsoft.com/en-US/vssetup/thread/60424309-bd78-4ca2-b618-03c4a16123b6
    Monday, January 19, 2009 3:55 AM
  • Yes, it can be restarted, and cluster resources can be moved back to the node without any indications that something went wrong before.

    Here is a part of cluster.log highlighting that:

    0000b5c.00000ba0::2009/01/14-13:50:23.028 INFO [Qfs] GetDiskFreeSpaceEx Q:\MSCS\, status 0
    00000b5c.00000b60::2009/01/14-13:52:33.903 INFO [CS] Received service stop command
    00000b5c.00000b6c::2009/01/14-13:52:33.918 WARN [INIT] The cluster service is shutting down.
    00000b5c.00000b6c::2009/01/14-13:52:33.934 INFO [EVT] EvShutdown
    00000b5c.00000b6c::2009/01/14-13:52:34.028 WARN [FM] Shutdown: Failover Manager requested to shutdown groups.
    etc.

    That is, everything indicates that the cluster service has been given (by something) a stop command, and has shutdown in an orderly fashion.

    Thank you
    Monday, January 19, 2009 7:32 AM