Time Limits on jobs

DSguru2B · Post by **DSguru2B** » Thu Dec 14, 2006 2:00 pm

Dont even think about killing the process using the dreaded 'kill' command. There can be open connections to a database, runaway processes that would leach over the resouces, not to mention, you getting a call from system admin.
Try my method, set the warning level to 1, and if your timer times out, just log a warning message, this will force the job to abort.

NEO · Post by **NEO** » Thu Dec 14, 2006 2:18 pm

Coming to think of it, we use exceptional handlers connected to an abort stage to abort the master sequencers. So to abort jobs, using your dsjob -log -warn might actually invoke the exception handler and abort the job. I will have to try that.

NEO · Post by **NEO** » Thu Dec 14, 2006 2:37 pm

Nope. That doesnt work. The sequencer continues to run inspite of the exception handler. I think I will have to go with the zero warning approach

DSguru2B · Post by **DSguru2B** » Thu Dec 14, 2006 2:43 pm

Exception handler only fires if anything goes wrong in the sequence job and it is not handled. Thats when the Exception handler fires.

NEO · Post by **NEO** » Thu Dec 14, 2006 3:09 pm

Even invoking the master sequence with dsjob warning limit of 1 and then logging a warning after timeout didnt work.

Code: Select all

$dshome/bin/dsjob -run -jobstatus -warn 1 $DSproject $DS_job_name

The master sequence job should abort but it didnt??!![/code]

NEO · Post by **NEO** » Thu Dec 14, 2006 4:10 pm

I am giving up hope. Doesnt seem like there is a clean and elegant way to abort a job from the comand line. I will have to go with my workaround plan of masking my master sequence in a job activity and use my wait for done file approach. That works, but I will have to create these timer super sequencers to run master sequencers. any other ideas??

ray.wurlod · Post by **ray.wurlod** » Thu Dec 14, 2006 7:48 pm

If you can specify under exactly what conditions you want an abort to occur, then you could create another job (Job Control code) that polled for those conditions and issued appropriate DSStopJob() calls.

Have you investigated a Terminator activity in a job sequence? That does much the same thing, but has to be triggered.

You can force the Exeception Handler to trigger only by guaranteeing that the sub-jobs abort. But you could set up user variables or routines that detect the exit status of all sub-jobs and trigger a control stream that ends with a Terminator.