[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Condor-users] condor_q -global stressing our schedulers?

Ian, I've found condor_status -submitters to be a useful alternative to
condor_q -global. It doesn't give all the job detail for each submitter
but does a nice job of listing all the submitting users and their total
job status (like number of jobs idle, running, held, flocked, etc.)
Since it gets info from the collector it is less stressful on the
schedd's. ~Brooklin 

-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx
[mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Ian Chesal
Sent: Friday, February 04, 2005 12:30 PM
To: Condor-Users Mail List
Subject: [Condor-users] condor_q -global stressing our schedulers?

I'm getting a lot of timeouts from our schedd machines if I call them
directly with condor_q and I have a sneaky suspicion that it's due to a
fair number of users of our system calling condor_q -global in scripts
that parse the output to display overall system state.

Is it possible for condor_q -global to stress schedulers to the point
where calls for queue status start getting dropped? I'm even seeing just
"condor_q" on the larger schedd machines (machines with 1000+ jobs
queued) issue "failed to fetch ads" messages. This is all with 6.7.3 on
a mix of Windows and Linux machines.

Is there a recommended (non-stressful) way for me to guide my users
towards so they can see who's has what running, queued and held in the
system and what the JobPrio of those jobs are?

- Ian

Ian R. Chesal <ichesal@xxxxxxxxxx>
Senior Software Engineer

Altera Corporation
Toronto Technology Center
Tel: (416) 926-8300

Condor-users mailing list