User Tools

Site Tools


batch

This is an old revision of the document!


Batch Jobs

A batch job is simply a set of commands grouped in a single file which are to be executed on a node or set of nodes assigned to the job by the scheduler. Below is an example batch job which contains a single command hostname. This command prints out the name of the host running the command.

tres-l1:pwolinsk:$ cat hostname.pbs 
hostname
tres-l1:pwolinsk:$ 

You can use the qsub command to submit this job to the queue. Without any additional parameters to the qsub command the scheduler will select the default parameters (i.e. default queue, walltime, number of nodes and number of cores).

tres-l1:pwolinsk:$ qsub hostname.pbs
283319.torque

When the job finishes, the output and error messages from the job will be written to new files ending in .o<JOBID> for output and .e<JOBID> for error messages.

tres-l1:pwolinsk:$ ls -ltr |tail -3
-rw-rw-r--   1 pwolinsk pwolinsk       9 Sep 19 10:16 hostname.pbs
-rw-------   1 pwolinsk pwolinsk     266 Sep 19 10:18 hostname.pbs.o283319
-rw-------   1 pwolinsk pwolinsk       0 Sep 19 10:18 hostname.pbs.e283319
tres-l1:pwolinsk:$ cat hostname.pbs.o283319 
tres1005
PBS Job Statistics:

PBS Input:
hostname

PBS Job ID: 283319.torque
Resource List: walltime=00:30:00
Resources Used: cput=00:00:00,mem=0kb,vmem=0kb,walltime=00:00:00
Queue Name: q30m32c
Program Return Code: 0
Head Node:
tres1005
nnodes= 1 ncores= 1 njobs=1
tres-l1:pwolinsk:$ 

The output in hostname.pbs.o283319 shows the output of the hostname command followed by the batch job scipt and accounting information from the scheduler. It shows that the job was submitted with a 30 minute walltime into queue q30m32c, and 1 core on 1 node was requested. To specify different values for these parameters we you can pass flags to the qsub command:

tres-l1:pwolinsk:$ qsub -q q06h32c -l walltime=1:00 -l nodes=1:ppn=32 hostname.pbs
283320.torque

The command above submitted the same job to the q06h32c queue with a walltime of 1 minute requesting all 32 cores on 1 node.

batch.1474298980.txt.gz · Last modified: 2016/09/19 15:29 by pwolinsk