datastage job time limit

randy · Post by **randy** » Tue Mar 21, 2006 9:53 am

Can someone suggest a method for limiting how long a job can run and aborting it if it exceds the time limit?

I have a job that usually runs in less than a minute, but occasionally hangs and will run forever. Since no errors are being generated, troubleshooting it is hard. So until I stumble on why this is happening, I need to abort the job when this happens.

Thanks
Randy

kcbland · Post by **kcbland** » Tue Mar 21, 2006 9:55 am

Custom job control that monitors when it starts a job, how long jobs are expected to run, and acts accordingly.

ArndW · Post by **ArndW** » Tue Mar 21, 2006 9:58 am

Does the job actually terminate if you send it a STOP request from the Director? If that works then you could write another job that waits (n) minutes, checks to see if the status of the other job is still RUNNING and issues a STOP request to it.

If a STOP doesn't work then it becomes more complicated. I think that solving the cause for the hang is more important. Do you know what the cause for the condition is and what phase/portion of the job the process is hanging on?

randy · Post by **randy** » Tue Mar 21, 2006 10:09 am

the job is real simple. It's an ODBC to hash file. The hash file is then used in many jobs downstream.

there are no errors in the log. It just hangs.

It has happend 4 times in 3 months and run successfully about 1000.

kcbland · Post by **kcbland** » Tue Mar 21, 2006 10:11 am

Any chance you're exceeded 32bit limitations?

randy · Post by **randy** » Tue Mar 21, 2006 10:17 am

You got me on this 32bit limit thing. How would I know?

ArndW · Post by **ArndW** » Tue Mar 21, 2006 10:29 am

32Bit means exceeding a 2Gb limitation on file sizes; this applies to your Hashed file. I would have expected DS to abort with a write error, but it could also cause the job to abort in the background but still show a status of "running". Have you been able to see if the job "hangs" after loading the records to the hashed file? If and when it hangs again, what does the job monitor show and can you see if the process is actually still running at UNIX level.