If you identify the job name that is eating all the CPU, and if it is the same job name every time, you could
(a) kill the job & watch to see if it restarts, and kill it again
(b) setup something to monitor for that job, to kill it any time it tries to start, sending a message to the jobname owner that the software is flawed for this platform & needs to be fixed before we can let it run here ... include your contact info in the message
(c) wait for the job owner to get in touch with you

I have encountered similar scenarios where
(a) someone is doing something interactive that needs to be batch
(b) the software written very poorly from a performance perspective e.g. read every record in a file (or several files) to find the fraction of 1% the program really needs to access. Old style RPG used to do this a lot ... reading records in a loop, testing for whether the records coded for deletion, not relevant etc. when those control values could be in a logical that filters out what is not wanted before the records even get to the RPG program.

Yes, you're right; I may have to set my alarm for some ungodly hour;
won't be the first time for that.

This thread ...


Return to Archive home page | Return to MIDRANGE.COM home page