[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] IDLE jobs
- Date: Thu, 4 May 2006 08:14:24 +0300
- From: "Anton Kucherov" <anton-k@xxxxxxxxxx>
- Subject: Re: [Condor-users] IDLE jobs
Probably what happened was that Condor vacated the first job for some
reason. I assume that the new job had a higher priority or was ranked
higher by the scheduler.
If you want to prevent that you should change the WANT_VACATE config
value at your configuration.
[mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Hrant P.
Sent: Thursday, May 04, 2006 6:02 AM
To: Condor User's
Cc: Glen Ferguson
Subject: [Condor-users] IDLE jobs
We're experiencing something somewhat odd with Condor. Presently,
all of our nodes are for production only (i.e., none of the machines
are used as workstations) and the START ClassAd on all machines is
statically set to TRUE. The machines all have 2 processors and we
haven't tinkered with the virtual machine settings, so the number of
VM's per node is 2.
Here's the actual problem...Two jobs were sent to node-5. The other
nodes in our cluster became saturated with jobs and the Condor
scheduler somehow sent another job to node-5. When this happened the
status of the first job started on node-5 became "I" and the new job
had a status of "R". As I understand things, once the scheduler
started the two jobs on node-5 a third job should NOT have been
allowed at all since there are only 2 virtual machines. What am I
Hrant P. Hratchian, Ph.D.
E. R. Davidson Fellow
Department of Chemistry
Bloomington, Indiana 47405-7102
"Liberty without learning is always in peril; learning without
liberty is always in vain."
John F. Kennedy
Condor-users mailing list