[torqueusers] job arrays and routed queues

Glen Beane glen.beane at gmail.com
Tue Jul 15 09:26:14 MDT 2008


On Tue, Jul 15, 2008 at 4:16 AM, Stijn De Weirdt <Stijn.DeWeirdt at ugent.be>
wrote:

> hi all,
>
> i'm trying out a routed queue setup with queue A on machine X forwarding
> jobs to queue B on machine Y.
>
> this cuurently works for simple single jobs (eg qsub -q A job1.sh), but
> submitting a job array (even a single job) fails.
>
> qsub -q A -t 0 job1.sh
> gives in tracejob
> ...
> Job: 55-0.X.site
> send of job to B at Y.site failed
>                          error = 15003
> Job rejected by all possible destinations
> ...
>
> on Y i increased the loglevel to 5, but nothing useful (nothing even
> indicating that there was an attempt).
>
> what is so differnet between an array of 1 job and a single job that this
> fails?



internally there are different steps that need to be taken when a job array
is being queued.  These happen even for an array that only consists of a
single job.   Job arrays are still under development, and had not been
tested in routing queues yet.  I will add this onto my todo list, but it
will probably be a couple weeks before I can come back to working on job
arrays
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.supercluster.org/pipermail/torqueusers/attachments/20080715/63d71d1f/attachment.html


More information about the torqueusers mailing list