Re: [condor-users] Jobs Stop Migrating

[Joel Hernandez <joelh@xxxxxxxxx>]

Actually, since upgrading duey to Condor 6.6 from 6.4.7 (to match what 
is on louie), this problem has not reoccurred.

However since then, I haven't been able to run the simple MPI job listed 
in section 2.10.2 MPI Job Submission of the Condor manual.

Thanks,
Joel

Dan Bradley wrote:

> Joel Hernandez wrote:
>
> > We have two clusters, louie  and duey. Users submit their jobs on the 
> > louie cluster.  When all the nodes on louie are busy, the jobs flock 
> > to the duey cluster.  This works fine for three or four hours and 
> > then stops all together for several hours even though many runnable 
> > jobs are still in the queue.
> > The jobs start flocking again after several hours or immediately 
> > after a condor_restart is performed on louie.  However, after several 
> > hours all the jobs stop migrating again.  Has anyone had this problem? 
>
>
>
> Very odd.  When you say that you do a condor_restart on louie, what 
> daemons are running on the machine in question?  Are you restarting 
> the schedd, or is it just the collector and negotiator?
>
> In the schedd logs, you should see statements about the "flock 
> level".  Can you please check what this is doing during the time when 
> flocking is not working?
>
> Dan Bradley
> University of Wisconsin, Condor Project
>
>
> Condor Support Information:
> http://www.cs.wisc.edu/condor/condor-support/
> To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
> unsubscribe condor-users <your_email_address>


Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>