Re: [Condor-users] Match-making delays/errors in Condor-G

[Dan Bradley <dan@xxxxxxxxxxxx>]

Todd may be able to correct me or explain why, but my guess is that you 
can work around this problem with the following configuration setting 
for the negotiator:

NEGOTIATOR_MATCHLIST_CACHING = False

--Dan

Jan Ploski wrote:

> Hello,
>
> I have a problem with match-making not working properly/being too slow in 
> Condor 7.0.0. The scenario is as follows: I submit 120 Condor-G jobs 
> together from a script. They all begin in status Idle (as expected). I 
> then repeatedly invoke condor_q to observe changes in JobStatus and 
> LastRejMatchReason of all jobs as time progresses. What I see is that in 
> the increasing order of ClusterIds every jobs hits LastRejMatchReason == 
> "no match found" once. When this happens, the match-making cycle is 
> immediately aborted, so that all jobs with a greater ClusterId remain 
> Idle. In one of the following cycles the job which previously had trouble 
> is matched correctly, but then the job with ClusterId+1 runs into the same 
> "no match found" problem. So it takes at least as many match-making cycles 
> as there are jobs in the queue to get every one matched to the target 
> resource (which is the same for all jobs and does not constraint the 
> accepted job count in any way). Here is an excerpt from NegotiatorLog:
>
> 4/25 09:53:28     Getting reply from schedd ...
> 4/25 09:53:28     Got JOB_INFO command; getting classad/eom
> 4/25 09:53:28     Request 20658.00000:
> 4/25 09:53:28 Attempting to use cached MatchList: Failed (MatchList 
> length: 0, Autocluster: 0, Schedd Name: jploski@xxxxxxxxxxxxxxxxxxxxxx, 
> Schedd Address: <134.106.52.210:20346>)
> 4/25 09:53:28       Rejected 20658.0 jploski@xxxxxxxxxxxxxxxxxxxxxx 
> <134.106.52.210:20346>: no match found
> 4/25 09:53:28     Sending SEND_JOB_INFO/eom
> 4/25 09:53:28     Getting reply from schedd ...
> 4/25 09:53:28     Got NO_MORE_JOBS;  done negotiating
> 4/25 09:53:28   Schedd jploski@xxxxxxxxxxxxxxxxxxxxxx got all it wants; 
> removing it.
>
> Can anyone explain why this is happening?
>
> Regards,
> Jan Ploski
>
> --
> Dipl.-Inform. (FH) Jan Ploski
> OFFIS
> Betriebliches Informationsmanagement
> Escherweg 2  - 26121 Oldenburg - Germany
> Fon: +49 441 9722 - 184 Fax: +49 441 9722 - 202
> E-Mail: Jan.Ploski@xxxxxxxx - URL: http://www.offis.de
> _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>
> The archives can be found at: 
> https://lists.cs.wisc.edu/archive/condor-users/
>