Conference vaxaxp::vmsnotes

189.0. "ALP system hang during SHUTDOWN" by CSC32::D_BROWN (Dave Brown CSC-VSG/INTDRV) Wed Feb 12 1997 15:41

    
    
    	I'm working with a customer who has approximately 10 Alphas in a
    cluster, each with their own system disk. The VMS versions in the
    cluster are 6.1, 6.2, and 6.2-1H3.
    
    	Often, a system will hang during shutdown when dismounting disks
    served by other cluster members. The process running shutdown is
    stalled waiting on a DMT$_ lock on the volume it is trying to dismount.
    The DMT$_ resource shows no locks in the Granted or Conversion queues,
    just the one for the shutdown process in the Waiting queue. The process
    copy of the lock is on another cluster member.
    
    	The customer had had a complement of incompatible patches on his
    various system disks and that has been straitened out. He has good
    versions of ALPSHAD, ALPSCSI, ALPDISM, and ALPLAVC and still the
    problem persists.
    
    	Any Ideas?
    
    	Thanks,
    
    	Dave 

    re .0
    
    (I would check the locks on the problem resource on all machines.)
    
    Crash system when hung and escalate via IPMT.
    
    Ten system disks, huh? Yuk! Are you sure that they have a compatible
    set of patches and the same patches on all machines? How can you be
    sure of that?
    
    To simplify matters you might try to see whether the problem can be
    reproduced with just two systems i.e. shut down entire cluster and bring
    up two systems and then shut down one machine. (Temporary changes to
    the votes may be needed to allow this.) You may need crash dumps from
    both machines.
    

    There are a number of locking issues with MOUNT and DISMOUNT that are
    addressed in the ALPCOMPAT_062/ALPCLUSIO01_062 kits.  The CLUSIO01 kits
    are in the TIMA review process now, and should be available shortly.
    
    Cheers,
    jeff