Topic
3 replies Latest Post - ‏2010-09-20T15:14:30Z by RobertHaig
JonathanAllan
JonathanAllan
14 Posts
ACCEPTED ANSWER

Pinned topic System Alert message - Huh?

‏2010-09-13T13:48:51Z |
Anyone:

I'm getting this message sent from BF as an e-mail:

"An alert message has been generated by the Build Forge Management Console running on (redacted console server uri). The user associated with this message is System (redacted e-mail address). The alert text is: Failed to start build process for id [3] [P]."

It seems to be related to the start of one of my purge jobs, but I haven't figured out a way to verify that. All I know is the time of the messages matches the time of the purge and nothing else appears to be running then.

BF sends this alert 88 times over a five minute period, then spits out three messages like this one:

"An alert message has been generated by the Build Forge Management Console running on (redacted console server uri). The user associated with this message is System (redacted e-mail address). The alert text is: Purge of :/u/build_forge/mayo_builds/CMTT_Mainframe_test/CMTT_32 incomplete."

Any clues what this means? And how to stop either/both these from showing up?
Updated on 2010-09-20T15:14:30Z at 2010-09-20T15:14:30Z by RobertHaig
  • RobertHaig
    RobertHaig
    55 Posts
    ACCEPTED ANSWER

    Re: System Alert message - Huh?

    ‏2010-09-13T16:39:11Z  in response to JonathanAllan
    The console process is trying to start purges of jobs that need to be purged. It is failing to start these processes. Likely due to some system reason. Check your system logs. One possibility is a limit on the number of processes allowed for a user to run. Either through ulimit, or system settings (AIX limits this through settings in the ODM accessible by an admin through SMIT).

    Rather than just stopping them from showing up, you want to solve the system level problem prohibiting processes from starting.

    likely culprits are memory, file handles, or some system level limitation on processes. If you or your sysadmin can't find it, open a PMR.
    • JonathanAllan
      JonathanAllan
      14 Posts
      ACCEPTED ANSWER

      Re: System Alert message - Huh?

      ‏2010-09-20T14:10:53Z  in response to RobertHaig
      It's a Windows machine, though the DB is on an Aix box running UDB. The server logs don't show much - nothing was logged near the dates/times in the messages. I've checked the BF logs (anything with .log as the extension in the BF install directory tree) and the system event logs (all 4). Anywhere else I should look?

      Max Console Procs is set to 20 - should I be looking for a bigger or smaller #?

      Based on the timing, this is an auto-purge of my scratch jobs (>30 days old, not production). It's possible this purge is conflicting with the deletion of production jobs (>120 days old). How long should it take a job to purge?

      Thanks for any ideas!
      Jonathan
      • RobertHaig
        RobertHaig
        55 Posts
        ACCEPTED ANSWER

        Re: System Alert message - Huh?

        ‏2010-09-20T15:14:30Z  in response to JonathanAllan
        if you have a bunch of purges running, they will take up your max console proc slots. In the next release (I believe there is a hotfix to 7.1.1.4 with this feature too) there is a setting to limit the max number of purges (to keep this case from happening).

        if you believe it is your purges causing this (it may be) I suggest opening a PMR to get the latest 7.1.1.4 hotfix, and then upgrade to 7.1.2 when it comes out later this year.