SLURM and HPC¶
trasgu can run on SLURM clusters through Snakemake profiles.
See Snakemake profiles for example profile files and how to use them with a run directory.
Install SLURM support¶
The SLURM executor plugin is optional:
uv sync --frozen --extra slurm
Packaged SLURM profile¶
From a run directory:
trasgu_run --profile slurm
This resolves the SLURM profile packaged with trasgu.
Custom Snakemake profile¶
trasgu_run --profile /path/to/snakemake-profile
Extra Snakemake arguments can be appended:
trasgu_run --profile /path/to/snakemake-profile --printshellcmds
Before submitting¶
Run a dry run:
trasgu_run --profile slurm --dry-run
Check that:
trasgu.yamlpoints to accessible input data.- the chunk work directory (
.trasgu_<run>by default, oroutput_dirif configured) is on storage visible to compute nodes. trasgu_urlis reachable from compute nodes, or points to a local Zarr copy.- The requested variable size exists in the Chimera Zarr store.