Re: Looking for advice: How to detect a job is actually doing work
Charles Wilt <[email protected]>
| Newsgroups | gmane.comp.lang.as400.java |
|---|---|
| Message-ID | <CAJ=Tnc6gmG0W1pqkrLia3mDJowLr0LKtszrHnuw=uBJn4iCKZw@mail.gmail.com> |
On Thu, Aug 5, 2021 at 8:27 AM David Gibbs via JAVA400-L < [email protected]> wrote: > Folks: > > I'm looking for advice on the best way to detect if a job is actually > doing work. > > If the current call stack is different from the previous call stack, I'm > taking that as a hint that the job is doing work. > > Are there any other job attributes, or techniques, you can think of that > would be useful in identifying a hung vs. active job? > > How often do you see a "hung" job? I don't think I've seen one in the 25yrs I've been on the platform.. I have however seen lots of jobs stuck in a poorly written loop. Checking the call stack isn't usually enough to determine if that's what's actually happening as it does usually change...but we do have a lot of ILE procedures. With OPM, I suppose only the statement # is prone to change. Our monitoring looks for "high" (> 5% or so) CPU use over an extended period of 60min or so I think. But even then, it usually is up to the on-call person to debug the job and see if it's looping or working as designed. Charles -- This is the Java Programming on and around the IBM i (JAVA400-L) mailing list To post a message email: [email protected] To subscribe, unsubscribe, or change list options, visit: https://lists.midrange.com/mailman/listinfo/java400-l or email: JAVA400-L-request-+hD5IHI5Xscn3HwCXmMcX9BPR1lH4CV8@public.gmane.org Before posting, please take a moment to review the archives at https://archive.midrange.com/java400-l. Help support midrange.com by shopping at amazon.com with our affiliate link: https://amazon.midrange.com