Running jobs in Torque#
VSC clusters using the Torque job scheduler:
Other clusters might use the Slurm job scheduler
The workflow in the HPC is straightforward:
Create a job script
Submit it as a job to the scheduler
Wait for the computation to run and finish
A job script is essentially a Bash script, augmented with information for the
scheduler. As an example, consider a file
hello_world.pbs as below.
1#!/usr/bin/env bash 2 3#PBS -l nodes=1:ppn=1 4#PBS -l walltime=00:05:00 5#PBS -l pmem=1gb 6 7cd $PBS_O_WORKDIR 8 9module purge 10module load Python/3.7.2-foss-2018a 11 12python hello_world.py
We discuss this script line by line.
Line 1 is a she-bang that indicates that this is a Bash script.
Lines 3-5 inform the scheduler about the resources required by this job.
It requires a single node (
nodes=1), and a single core (
ppn=1) on that node.
It will run for at most 5 minutes (
It will use at most 1 GB of RAM (
Line 7 changes the working directory to the directory in which the job will be submitted (that will be the value of the
$PBS_O_WORKDIRenvironment variable when the job runs).
Lines 9 and 10 set up the environment by loading the appropriate modules.
Line 12 performs the actual computation, i.e., running a Python script.
Every job script has the same basic structure.
Although you can use any file name extension you want, it is good practice
.pbs since that allows support staff to easily identify your
More information is available on:
- Job resources
- Job names, output files and notifications
- Starting programs
Using the module system
Submitting and monitoring a job#
Once you have created your job script, and transferred all required input data if necessary, you can submit your job to the scheduler
$ qsub hello_world.pbs 205814.leibniz
qsub returns a job ID, an unique identifier that you can use to manage
your job. Only the number, i.e.,
205814 is significant.
Once submitted, you can monitor the status of your job using the
$ qstat Job ID Name User Time Use S Queue ------------------------- ---------------- --------------- -------- - ----- 205814.leibniz hello_world.pbs vsc30140 0 Q q1h
The status of your job is given in the
S column. The most common values are
job is queued, i.e., waiting to be executed
job is running
job is completed, i.e., finished.
More information is available on
By default, the output of your job is saved to two files.
This file contains all text written to standard output, as well as some information about your job.
This file contains all text written to standard error, if any. If your job fails, or doesn’t produce the expected output, this is the first place to look.
For instance, for the running example, the output file would be
hello_world.pbs.o205814 and contains
1===== start of prologue ===== 2Date : Mon Aug 5 14:50:28 CEST 2019 3Job ID : 205814 4Job Name : hello_world.pbs 5User ID : vsc30140 6Group ID : vsc30140 7Queue Name : q1h 8Resource List : walltime=00:05:00,nodes=1:ppn=1,neednodes=1:ppn=1 9===== end of prologue ======= 10 11hello world! 12 13===== start of epilogue ===== 14Date : Mon Aug 5 14:50:29 CEST 2019 15Session ID : 21768 16Resources Used : cput=00:00:00,vmem=0kb,walltime=00:00:02,mem=0kb,energy_used=0 17Allocated Nodes : r3c08cn1.leibniz 18Job Exit Code : 0 19===== end of epilogue =======
Lines 1 through 10 are written by the prologue, i.e., the administrative script that runs before your job script. Similarly, lines 12 though 19 are written by the epilogue, i.e., the administrative script that runs after your job script.
Line 11 is the actual output of your job script.
The format of the output file differs slightly from cluster to cluster, although the overall structure is the same.
Monitoring memory and CPU usage of programs, which helps to find the right parameters to improve your specification of the job requirements.
worker quick start: To manage lots of small jobs on a cluster. The cluster scheduler isn’t meant to deal with tons of small jobs. Those create a lot of overhead, so it is better to bundle those jobs in larger sets.
The Checkpointing framework can be used to run programs that take longer than the maximum time allowed by the queue. It can break a long job in shorter jobs, saving the state at the end to automatically start the next job from the point where the previous job was interrupted.