job status

przytula · Post by **przytula** » Fri Oct 28, 2005 3:15 am

we missed an instruction in a routine and the job keeps running
we tried to stop the job : status = running
we did shutdown of engine and restarted the engine = status running
we tried as in other topic
COPY FROM RT_LOGnnn TO VOCLIB '//PURGE.SETTINGS' OVERWRITING
CLEAR.FILE RT_LOGnnn
COPY FROM VOCLIB TO RT_LOGnnn '//PURGE.SETTINGS' DELETING
with 162 as jobnbr (found from admin select)
same result
Is there any way to stop/abort a job in a clean way ?
Thanks for all help
Best Regards, Guy Przytula

ray.wurlod · Post by **ray.wurlod** » Fri Oct 28, 2005 3:38 am

The job is stopped.

But it never got a chance to update its records in the RT_STATUSnnn table.

Probably quickest is to Clear Status File from Director.

Otherwise, you must reset the status value for job, resource and active stage records in RT_STATUSnnn, to explain which would take more time than I have right now.

Your clearing of the RT_LOGnnn table would have no effect, as you reported. Well done for thinking to preserve the purge settings, however.

przytula · Post by **przytula** » Fri Oct 28, 2005 3:54 am

thanks for the update/info
In director : job : stop is active but reset is gray = not selectable
we tried stop = server was informed, but did not change the status.
Any other way to reset the status (maybe on engine machine ?)
Thanks and Best Regards, Guy Przytula

kcbland · Post by **kcbland** » Fri Oct 28, 2005 6:01 am

The ONLY method you should use to stop a job in an emergency is the unix kill command. You are lucky you didn't corrupt your DS project with the things you did. NEVER attempt those things unless you absolutely understand the internals. You are lucky you didn't really screw things up.

To stop a job, do a "ps -ef |grep phantom" and look for your job. THE DSD.RUN is the parent process, the DSD.StageRun are the active stages. Issue a kill processid for one of the Stage run phantoms. Usually killing one of them is sufficient to make the whole job abort. But, you can kill them all including the DSD.RUN phantom. Since nothing's left to update the jobs status file, you can now simply recompile the job, clear the status using the Director, whatever. You can't RESET because the job still indicates running so the Director doesn't allow you.