[Condor-users] Waiting vanilla jobs in the queue when machines are available

[David Lalonde <davidl@xxxxxxxxxxxxxx>]

Hello


Sometime, some users complaint that their jobs are waiting even if the pool is empty or when some machines are available.

These persons are submitting vanilla jobs with around 600 sub-jobs  and they are very quick, let's say 5min each sub-jobs.

When almost all sub-jobs are done, the last sub-jobs are waiting maybe 5min and then they are executed. I would like to understand why Condor do this and if there is a way to disable this configuration in order to avoid the waiting time.

As far as I remember, this happening from the beginning, when we had the basic configuration.

I printed out a part of the Negotiationlog below:

04/28 14:25:21       Matched 15.82 STUDIO_compositing.davidla@xxxxxxxxxxxxxx <192.168.24.77:35569> preempting none <192.168.25.102:42735> slot1@xxxxxxxxxxxxxxxxxxxxxxxxxx
04/28 14:25:21       Successfully matched with slot1@xxxxxxxxxxxxxxxxxxxxxxxxxx
04/28 14:25:21     Request 00015.00083:
04/28 14:25:21       Rejected 15.83 STUDIO_compositing.davidla@xxxxxxxxxxxxxx <192.168.24.77:35569>: fair share exceeded
04/28 14:25:21     Got NO_MORE_JOBS;  done negotiating
04/28 14:25:21 Phase 4.3:  Negotiating with schedds ...
04/28 14:25:21   Negotiating with STUDIO_compositing.davidla@xxxxxxxxxxxxxx at <192.168.24.77:35569>
04/28 14:25:21 1 seconds so far
04/28 14:25:21     Request 00015.00083:
04/28 14:25:21       Matched 15.83 STUDIO_compositing.davidla@xxxxxxxxxxxxxx <192.168.24.77:35569> preempting none <192.168.25.111:59497> slot1@xxxxxxxxxxxxxxxxxxxxxxxxxx
04/28 14:25:21       Successfully matched with slot1@xxxxxxxxxxxxxxxxxxxxxxxxxx
04/28 14:25:21     Reached submitter resource limit: 1.000000 ... stopping

Any idea?

-- 
David Lalonde, Rendering Lead
Lumière VFX
Email: davidl@xxxxxxxxxxxxxx 
Phone: +1-514-316-1080x2049
Cell: +1-514-941-7448