Conference decwet::winnt-clusters

862.0. "Changing Cluster IP addresses" by ACISS2::DATZMAN (Vee Vont To Pomp You Up) Thu Jun 05 1997 20:29

    I have a customer with a fully functioning NT Cluster with the following
    atributes:
    
    NT 4.0, NT Service PAK 3
    Clusters 1.1 no service pak
    Cluster nodes are AlphaServer 4000 5/400s wiht 512 MB memory
    RAIDArray 410
    Dual Redundant HSZ40 RAID controllers
    SRM Console 4.8-7
    AlphaBIOS 5.28-1
    NT HAL 4.0B
    Dedicated 10/100 Ethernet card for cluster communications (DEC DE500)
    FDDI Controller (DEFPA) for normal network communicaitons
    
    The customer wanted to switch the role of the network cards.  That is,
    use the 10/100 Ethernet card for normal network use and use the FDDI
    card for cluster communications.  This was done because the customer was
    moving to switched ethernet for this building and wanted to use the
    10/100 Ethernet cards.
    
    Apparently they switched the IP addresses of the cards on one machine
    and then rebooted.  The system would not reboot and they received a
    K_MODE_EXCEPTION error.
    
    I do not know if they performed a manual failover before changing the
    address or if they stopped the cluster services, etc.
    
    Can anyone tell me if this scenario violates any rules of cluster
    configuration integrity?  Also as a bit of history these two cluster
    machines have been exposed to extreme heat (approx 120 degrees for an
    unknow period of time estimated from 2 hours to 24 hours).  Some of the
    hardware component have been replaced.
    
    Thanks,
    
    Dick