VCS shuts down the instance after it starts and issues resource faults

Use hagrp -clear to clear the resource faults. This command clears the resource faults for an VCS group named “asegrp1”:

hagrp -clear asegrp1

This error may result in the instance writing a number of errors to the error log:

• Failed to identify instance – VCS is using the wrong name to start the instance. To rectify:

  • Verify that the each RUN_server_name file includes the correct path to the quorum device, and that the value for server_name is the same as the name of the instance.

  • Use qrmutil to validate the quorum configuration, specifically the instance and host names:

    qrmutil -Qquorum_path --display=all
    

• Shutdown with nowait detected – VCS cannot determine if the instance is running. VCS uses the UNIX ps command with the server name in the execution string to verify whether an instance is running. To make sure the ps command can find each instance:

  • (Solaris only) Verify that the instance name in the RUN_server_name file appears within the first 80 characters

  • Use hares -display to verify that the Server attributes for the ASE resource. The Server attributes must correspond to the instance names. This example displays the instance names for the “aseres1” resource:

    hares -display aseres1 -attribute Server
    

    If necessary, use hares -modify to change the values:

    hares -modify aseres1 Server ase1 -sys hpcblade1
    

    If an instance requires a long time to start, Veritas may stop the dataserver process during startup and issue a resource fault. To prevent these resource faults, allow instances more time for startup by adjusting the following Sybase VCS monitoring parameters:

    • Adjust the time between monitor checks for the ASE resource. For example:

      hatype -modify Sybase MonitorInterval 60
      

    • Adjust the number of monitor checks to confirm the ASE resource is online. For example:

      hatype -modify Sybase OnlineWaitLimit 10
      

    • Adjust the number of failed monitor checks before issuing a resource fault for the Sybase resource. For example:

      hatype -modify Sybase ToleranceLimit 2
      

• Engine exited with signal 11 – the instance fails on the Linux platform. To rectify, reset the LD_POINTER_GUARD environment variable in the RUN_server_name files:

  • On the SUSE platform, enter:

    export LD_POINTER_GUARD=1
    

  • On the RedHat platform issue:

    export LD_POINTER_GUARD=0