condor_ vacate_job [-pool centralmanagerhostname[:portnumber] | -name scheddname ]| [-addr "<a.b.c.d:port>"] [-fast] cluster... | cluster.process... | user... | -constraint expression ...
condor_ vacate_job [-pool centralmanagerhostname[:portnumber] | -name scheddname ]| [-addr "<a.b.c.d:port>"] [-fast] -all
condor_ vacate_job finds one or more jobs from the Condor job queue and vacates them from the host(s) where they are currently running. The jobs remain in the job queue and return to the idle state.
A job running under the standard universe will first produce a checkpoint and then the job will be killed. Condor will then restart the job somewhere else, using the checkpoint to continue from where it left off. A job running under any other universe will be sent a soft kill signal (SIGTERM by default, or whatever is defined as the SoftKillSig in the job ClassAd), and Condor will restart the job from the beginning somewhere else.
If the -fast option is used, the job(s) will be immediately killed, meaning that standard universe jobs will not be allowed to checkpoint, and the job will have to revert to the last checkpoint or start over from the beginning.
If the -name option is specified, the named condor_ schedd is targeted for processing. If the -addr option is used, the condor_ schedd at the given address is targeted for processing. Otherwise, the local condor_ schedd is targeted. The jobs to be vacated are identified by one or more job identifiers, as described below. For any given job, only the owner of the job or one of the queue super users (defined by the QUEUE_SUPER_USERS macro) can vacate the job.
Using condor_ vacate_job on jobs which are not currently running has no effect.
Do not confuse condor_ vacate_job with condor_ vacate. condor_ vacate is given a list of hosts to vacate, regardless of what jobs happen to be running on them. Only machine owners and administrators have permission to use condor_ vacate to evict jobs from a given host. condor_ vacate_job is given a list of job to vacate, regardless of which hosts they happen to be running on. Only the owner of the jobs or queue super users have permission to use condor_ vacate_job.
When vacating a PVM universe job, you should always vacate the entire job cluster. (In the PVM universe, each PVM job is assigned its own cluster number, and each machine class is assigned a ``process'' number in the job's cluster.) Vacating a subset of the machine classes for a PVM job is not supported using condor_ vacate_job. To vacate individual nodes in a PVM computation, you must use condor_ vacate and target the specific hosts.
To vacate job 23.0:
% condor_vacate_job 23.0
To vacate all jobs of a user named Mary:
% condor_vacate_job mary
To vacate all standard universe jobs owned by Mary:
% condor_vacate_job -constraint 'JobUniverse == 1 && Owner == "mary"'Note that the entire constraint, including the quotation marks, must be enclosed in single quote marks for most shells.
condor_ vacate_job will exit with a status value of 0 (zero) upon success, and it will exit with the value 1 (one) upon failure.
U.S. Government Rights Restrictions: Use, duplication, or disclosure by the U.S. Government is subject to restrictions as set forth in subparagraph (c)(1)(ii) of The Rights in Technical Data and Computer Software clause at DFARS 252.227-7013 or subparagraphs (c)(1) and (2) of Commercial Computer Software-Restricted Rights at 48 CFR 52.227-19, as applicable, Condor Team, Attention: Professor Miron Livny, 7367 Computer Sciences, 1210 W. Dayton St., Madison, WI 53706-1685, (608) 262-0856 or miron@cs.wisc.edu.
See the Condor Version 6.8.3 Manual for additional notices.