DX Unified Infrastructure Management

  • 1.  Clustered Application Monitoring

    Posted Feb 22, 2010 11:53 PM
    We have a few applications that are clustered via third party software, with current monitoring when application failover occurs disks, services/processes go into alarm since they are not running. Does anyone else have a similar setup?

    I was thinking some type of correlation LUA script but didn't know if there was an easier route.


  • 2.  Clustered Application Monitoring

    Posted Feb 23, 2010 12:14 AM
    I think you could use a NAS script or triggers to capture this type of issue.  If you setup monitoring of all of the disks, services, and processes on both hosts, you really only want to see an alarm if there are alarms from both hosts.  Then under normal circumstances, one of the hosts should always be generating alarms.  If you can mark the alarms as invisible with a pre-processing rule, you could use AO triggers to make new alarms if both hosts open the same alarm.  It would probably take several triggers to accomplish this, so a NAS script might be a bit simpler in the end.

    Not exactly simple overall, but I think it would get you the result you want.

    I would really love to see the cluster probe support a cluster type of "custom" that would allow you to provide one or more commands that the probe would use to determine the cluster status.  Then you could have your own scripts or cluster commands tell the probe when the failover the monitoring to the other node.

    -Keith