extras

Cluster status scripts

If snakemake is having trouble detecting job failures, you can provide a custom script to --cluster-generic-status-cmd. You can either do this at the command-line or add it to your config.v8+.yml.

Important: To use any of these, you must add the flag --parsable to the call to sbatch in the field cluster-generic-submit-cmd in config.v8+.yml

Important: These scripts must be executable: chmod +x <script>

Sources:

• Snakemake documentation for --cluster-generic-status-cmd
• slurm-status.py from official Slurm profile

Scripts:

• status-sacct.sh - Bash script that uses sacct to determine job status (recommended)

• status-sacct.py - Python script that uses sacct to determine job status

• status-scontrol.sh - Bash script that uses scontrol to determine job status

• status-sacct-multi.sh - Bash script that uses sacct to determine job status in a multi-cluster setup

• status-sacct-robust.sh - Bash script that re-runs sacct multiple times if it fails to return a valid status

I prefer to keep scripts as simple as possible. However, if none of the above simple scripts are sufficient for your use case, I recommend you try one of the following available status scripts:

• LUMC/slurm-cluster-status

• slurm-status.py from the official profile

0