[gridengine users] exit value when killed

[Rayson Ho]

On Thu, Jan 5, 2012 at 11:47 AM, Gerard Henry <ghenry at cmi.univ-mrs.fr> wrote:
> in 6.2u5, a job finished, but the program was killed (*) due to lack of
> ressources (no more ram):
> in sge err:
>   /local/export/sge/default/spool/charybde/job_scripts/14561: line 12: 25776
> Killed ./benchruntime

You may want to check the resource limits inside a job (run "limit"
for csh/tcsh or "ulimit -a" for bash/sh/ksh inside a job), and see if
they are different than the limits set by an interactive shell.


> I runned it without SGE, this program needs more ram than available on the
> host. But is it normal that qacct says that everything is ok?
> failed       0
> exit_status  0

Looks like your job script does not return the exit status of the
"benchruntime" program - as SGE waits for the job script, only the job
script's exit status is recorded by SGE.

The simplest way to return the exit status:

#!/bin/sh

benchruntime
exit $?

Rayson


>
>
> Does it mean that i can only interpret the non empty sge err file to tell
> that the program exits abnormally?
>
>
> thanks in advance,
>
> gerard
>
>
> (*) killed by the operating system, i guess?
> _______________________________________________
> users mailing list
> users at gridengine.org
> https://gridengine.org/mailman/listinfo/users