[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] MPI Jobs again.
- Date: Tue, 16 Aug 2005 09:45:09 -0500 (CDT)
- From: "Jeremy Mann" <jeremy@xxxxxxxxxxxxxxxxxxx>
- Subject: Re: [Condor-users] MPI Jobs again.
This is a followup question to the email below. What happens if you don't
have DedicatedScheduler when you run the condor_status command?
I'm having the same MPI problems. Jobs will submit, but won't execute,
they just sit in the queue idle.
Chaitanya V. Hazarey said:
> Hi All,
> Thanks very much for the reply.
> I got the problem solved. Its working now.
> Greg Thain wrote:
>> Chaitanya V. Hazarey wrote:
>>> Hi all,
>>> It would also be helpfull if any one could point out how to diagnose
>>> this situation. I have tried D_ALL and also D_DEBUG for schedd. But
>>> nothing intersting comes up. What sould I look for.
>> Most likely, the startds on your execute machine are not reporting as
>> willing to be scheduled by the DedicatedScheduler. To verify this, run
>> condor_status -l | grep Dedicated
>> For each machine, you should get two lines, that look like:
>> Rank = Scheduler =?= "DedicatedScheduler@xxxxxxx"
>> DedicatedScheduler = "DedicatedScheduler@xxxxxxx"
> What was wrong was that the hostname part of the DedicatedScheduler@
> contained the name of that machine not the dedicated scheduler. That was
> the problem.
> Chaitanya V. Hazarey
> Condor-users mailing list
University of Texas Health Science Center
Bioinformatics Core Facility
Phone: (210) 567-2672