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

[Jaime Frey <jfrey@xxxxxxxxxxx>]

On Apr 25, 2008, at 10:17 AM, Jan Ploski wrote:

> Maybe I was happy too soon: another problem seems to have been  
> introduced
> now. Even though the first 100 jobs or so got submitted, the  
> remaining 20
> now just remain Idle (JobStatus == 1 according to condor_q) and
> unsubmitted. NegotiatorLog repeatedly contains
>
> 4/25 17:12:46 Phase 4.1:  Negotiating with schedds ...
> 4/25 17:12:46     NumStartdAds = 3
> 4/25 17:12:46     NormalFactor = 1.000000
> 4/25 17:12:46     MaxPrioValue = 4.000000
> 4/25 17:12:46     NumScheddAds = 1
> 4/25 17:12:46   Negotiating with jploski@xxxxxxxxxxxxxxxxxxxxxx  
> skipped
> because no idle jobs
> 4/25 17:12:46   Schedd jploski@xxxxxxxxxxxxxxxxxxxxxx got all it  
> wants;
> removing it.
>
> What can I do to diagnose it?


It sounds you're running into Condor's default limit on the number of  
jobs to submit at a time to a given resource for grid universe jobs.  
Try setting GRIDMANAGER_MAX_SUBMITTED_JOBS_PER_RESOURCE to a number  
larger than 100 in the Condor config file.

Thanks and regards,
Jaime Frey
UW-Madison Condor Team