slurm-hpc — generic Slurm workload-manager know-how

Portable Slurm reference for authoring batch jobs and reasoning about resource isolation on any cluster. Not tied to a specific site — for a specific repo's recipes and layout, prefer that project's own skill.

When to use

• Writing or fixing an sbatch script, or an srun command line.
• Choosing resource requests (CPUs, memory, GPUs, time, partition).
• Reading job/queue/node state (squeue, sacct, sinfo, scontrol).
• Reasoning about what isolates jobs (cgroups) and what fences GPU VRAM.

When NOT to use

• Operating a specific project's cluster with its own recipes → use that repo's skill.
• Designing cluster provisioning/config from scratch → that's admin work, not this.

Batch script skeleton