On Thu, 8 Sep 2016, sjones wrote:
> Hm...
>
> LHCb is guaranteed a share of our ARC/Condor cluster, so it is not relevant
> if there are a lot of ATLAS jobs waiting. New LHCb jobs would run ahead of
> earlier ATLAS jobs if the LHCb share is not yet filled. So there is no use in
> considering ATLAS traffic.
Right, but that affects only already pending jobs, and not new
submissions, doesn't it?
>
> But at present we have many ATLAS jobs waiting, and very few LHCb ones.
> Perhaps this is the same thing that you have seen?
>
That could be. Do you know how your ARC reports the pending jobs to LHCb
(or in general)?
To see if it is the same thing like I see, then you could have a look to
when the last LHCb job that is still pending was submitted (if any) and
how many still pending ATLAS jobs were submitted before this one. If it is
the same, then there should be only a few ATLAS jobs that are
still pending submitted before the last LHCb job.
If there are more than about 20-50 pending ATLAS jobs submitted before the
last pending LHCb job, then it could be the same problem. If not, it would
be interesting to see how your ARC reports the pending jobs.
Cheers,
Marcus
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|