7 Replies Latest reply on Sep 27, 2017 10:23 AM by marcus.partridge

    alerting on process instances running too long


      We have occasional rampant "ghost" transcode processes on our Telestream Lightspeed servers that sit there at 2-5% CPU and do nothing. We'd like to get an alert when the oldest instance is running over NN hours.


      The thing is though to not alert if no instances are running. In others words:

      • zero transcode processes: good
      • individual transcode process instances running under NN hours: good
      • individual transcode process instances running over NN hours: send an alert


      The bottom line is to flag a condition when a specific process runs too long yet not to flag anything (e.g. not to mark a node as having an issue) when no instances are running.


      What are my options?


      Seems there are two ways:

      1. A solution suggested by mdriskell to "Is there way to monitor how long a process is running?": Process Run Time that uses PS. Only problem: it always reports "down" regardless of whether a process is running. I do have the prerequisites (PS 2.0 is installed) and haven't had a chance to troubleshoot this yet.
      2. A solution suggested by aLTeReGo on the same thread (Is there way to monitor how long a process is running?) using "Windows Performance Counter". This one works beautifully except when no process in question is running, I get "Node is Up. Application 'Process running too long?' has state: Unknown." and the node is basically considered by Solarwinds to have a "problem" when it doesn't. Is there a way around that?