[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Condor-users] job evicted but stays idle!


I've been using condor (v7.4, all windows machines running either WinServer2003 or WinXP) for a few months now, but am confused about one thing:

when a job gets evicted because of a user interacting with his machine, the job seems to permanently remain in an "idle" state, at least according to condor_q command. I waited about 30 minutes. I ran 

condor_q -analyze


condor_q -better-analyze

but it didn't tell me anything useful. It just said that 

<all machines> are rejected by your job's requirements

This did not make sense to me, since the requirements should have been fulfilled by one machine which was "available". (I'm 100% positive this machine had 4 slots available; i checked via condor_status; moreover, this one machine is a "server" machine dedicated to condor and so does not care whether there's user activity or CPU usage.)

Finaly, I removed the idel job via condor_rm and I resubmitted the rescue dag. The job then started running, as expected.

So, why is it that evicted jobs are remaining in this permanent idle state? Or, does anyone have any suggestions of commands I should run and/or log entries I should look at to uncover the problem? Anyone else seen this?

Thanks in advance, all. And happy holidays!