[Condor-users] unkillable held jobs

[Ilya Narsky <narsky@xxxxxxxxxxxxxxx>]

A user submitted a bunch of jobs to the globus universe. These jobs were
never executed and went into 'held' state. We successfully removed them
from the condor queue on the executing site but these jobs are still held
in the condor queue on the submitter site. When I try to remove them on
the submitter site by condor_rm, they go into X state and then a while
later back into H state. They have been like this for 5 days now and yet
they are still unkillable. Is there any other way to remove these jobs
from the queue? Is there an adjustable timeout somewhere in the condor
config on how long a job can be held in the queue?     Thanks,   Ilya