User Tools

Site Tools


tier3:condor

Condor Batch Scheduler

Performance

Direct reads/writes to your home directory on shell.tier3 (NFS mounted to the worker nodes) can cause performance problems.

The options for telling condor to copy files into and out of your job are:

  • Should_Transfer_Files = YES
  • Transfer_Input_Files = file1, file2
  • WhenToTransferOutput = ON_EXIT

Job Status

To see all your jobs and their status:

$ condor_q username 

Condor will hold your job when a problem occurs. Held jobs remain in the queue, waiting for user intervention. Once the problem is resolved, use condor_release to continue running the job.

Display jobs currently running, and on which machines

condor_q -r 

To see why your job, with id 855969.0, is not running:

$ condor_q -better-analyze 855969.0

To see a long description of job with id 855969.0:

$ condor_q -l 855969.0 | less

To see what machines are available in the cluster and their status:

$ condor_status

To see priority:

$ condor_userprio -allusers -all

View local universe/managed fork jobs

condor_q -constraint "JobUniverse == 12"

View non not local universe/managed fork jobs

condor_q -constraint "JobUniverse != 12"

View Globus-specific status

condor_q -globus

Common User Commands

Place job with id 855969.0 in the queue on hold

$ condor_hold 855969.0

Release job with id 855969.0 in the queue that is on hold

$ condor_release 855969.0

Submit a job

 $ condor_submit

Delete job with id 855969.0

condor_rm 855969.0 

To remove all your jobs:

$ condor_rm username

Administration

How to shut down a single execute node without killing jobs

condor_off -startd -peaceful <hostname>

HTCondor Wiki

tier3/condor.txt · Last modified: 2013/02/05 17:34 by tlknight