Subject: [Condor-users] submit jobs complete but then go back into queue
I have a simple submit job that executes
on a client via Condor, but then returns to the Condor pool. I think the
problem is related to the Condor daemon owners, the submit owner and the
ad owner. Our daemons run under a service account because we are restricting
the configuration files on a shared drive. I am not sure this is the best
approach yet, but we are still trying to meet all of our security concerns.
We are using a small pool for testing,
SSL, NTSSPI, All condor clients and CM is Windows XP as well (SP3). SSL
is used for client authentication. If I run the Condor job under the condor
service account (the same account that the condor daemons run as, the jobs
exit the Queue properly. If I run the Condor job as my user account, the
job executes, but then returns to the Queue as 'I', and eventually it will
run again and again and so forth.
I tried to specify the owner class add
in the submit file but when I do this the job sits in idle and never runs.
Does anyone have any ideas as to what I might be doing wrong or have any
recommendations as to how we should handle running the daemons under a
different account. What does not make sense to me is that the state changes
from idle to run after the job is submitted, but then can not exit the
queue after it completes when the class ad (odonnellm) and submit owner
(igskabcb-condoradmin) are different.
PS I do not see anything else out of
the norm in the other log files (submit or execute machines).
Thank you for your help/suggestions,
Mike
Below are snapshots of the log files
indicating the errors when the state of the job is trying to be changed.