Condor |
| |
The cluster currently uses a scheduling tool called condor (a full users manual can be found in this web page). This tool manages jobs by accepting a condor submission file via the condor_submit command. Two example files, “vasp.submit and “condor2.submit,” are located in the /group/nnin/condor directory on the cluster. VASP is, in particular, a parallel code and requires a separate MPI script in order to run. The MPI script is particular to the MPI installation (MPICH, LAM, openmpi) and is currently supported on this cluster for MPICH1 only. As you can see from the vasp.submit file, the script, which takes the place of the executable in the condor_submit file, is /opt/condor/bin/condor-mpirun.
Three useful condor commands are the following:
condor_submit submit_file_name # submits a job to the cluster, see above
condor_q # lists the jobs currently running on the cluster “condor_q your_userid” will give only your jobs,
condor_rm [your_userid or job_number] # removes all of your jobs or the specified ones (kills them if they have started) from the cluster. |
| |
| |