First you need to verify that your environment is properly set.

>ssh login@surf

The sbatch command allow you to submit a script with commands to the cluster

>sbatch -p surf cmd

• The command then gives a job execution number.
• >scontrol show job <job-number> allows you to follow the execution of the job
• >scancel <job-number> allows you to kill the job

The job will generate output files:

• slurm-<job-number>.out in the directory from which the sbatch command was executed
• slurm-<job-number>.e in case of errors

In case of necessity the debug line sbatch -p debug allows you to use two additional lames (16 procs with 24GB memory)

There are options to:

• specify the memory required >sbatch -p surf –mem=xxMB
• specify the number of processors required >sbatch -p surf –cpus–per-task=24 (if you specify more than 24, the number of available lames for your job will be very limited)

The cmd script must start with

#!/bin/bash

and should set the necessary environement. All alias can be checked with the command >alias