[prev in list] [next in list] [prev in thread] [next in thread] 

List:       grid-engine-users
Subject:    Re: [GE users] SGE 5.3 migration from RH 7.1 to Debian 3.0: Problems
From:       Andy Schwierskott <andy.schwierskott () sun ! com>
Date:       2003-01-29 11:02:25
Message-ID: Pine.GSO.4.44.0301291200190.22088-100000 () sr-ergb01-01
[Download RAW message or body]

Johannes,

what "failed" code do you see in qacct for that job? What stdout/err files
or created by this job? Is the queue after the job failure set to error
state?

Andy


On Mon, 27 Jan 2003, Johannes Graumann wrote:

> Hello,
>
> Don't know exactly where to start, so here it is:
>
> - was running SGE 5.3 as the queuing mechanism on a RH 7.1 system
> sucessfully. Job executed was 'qsub -S /bin/bash -cwd program_script', where
> 'program_script' was 'screen -D -m binary'. Worked seamlessly (once I had
> figured out that the stupid binary held on to STDIN and OUT and how to
> overcome the problem with screen - but that's another story).
>
> - since my personal preference has always been debian, I recently migrated
> the whole system to debian 3.0 (woody). Reiterated the sge setup (which had
> been documented).
> Now saying 'qsub -S /bin/bash -cwd experiment_script' where
> 'experiment_script' is
> 'ls -l /usr/bin > experiment.txt' works.
> But 'qsub -S /bin/bash -cwd program_script', where 'program_script' is
> 'screen -D -m binary' - which worked on RH7.1 - fails (gets in the queue,
> gets transfered, shuts down immediately - as followed by 'qstat'). Great
> confusion since just saying 'screen -D -m binary' works perfectly fine.
>
> Might this have to do with having switched to SGE5.3p2 rather than using
> SGE5.3 as in the beginning?
>
> Does anybody have any suspicion what might be going on here or at least
> where I can look why this is failing.
>
> Utterly puzzeled,


[prev in list] [next in list] [prev in thread] [next in thread]