Slurm and job scripts¶
How Slurm runs jobs on the compute nodes, and how you write, submit and follow your own.
| Page | What it covers |
|---|---|
| How Slurm works | the login node and the compute nodes, what a job asks for, partitions and the queue |
| Your first job | a job script line by line, from your computer to the cluster and back |
| R and Python in a job | R and Python from the modules of the cluster |
| Many runs: job arrays | one job script that runs once for each line of runs.txt |
| Interactive jobs | a terminal on a compute node, to try things out |
| Choosing time and memory | how to estimate what a job needs, and what happens when it asks for too little |
| Asking for a GPU | a job on a node with a GPU |