# slurm-workload-manager

> Build, configure, run, and test the Slurm Workload Manager from this repository. Trigger when asked to compile Slurm, generate slurm.conf/slurmdbd.conf, start daemons locally, submit or debug jobs, run the test suite, or answer how to use Slurm commands and plugins.

- Author: Siyu Liu
- Repository: IntelligentMat/Materials-Science-Skills-For-LLM
- Version: 20251221161123
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-07
- Source: https://github.com/IntelligentMat/Materials-Science-Skills-For-LLM
- Web: https://mule.run/skillshub/@@IntelligentMat/Materials-Science-Skills-For-LLM~slurm-workload-manager:20251221161123

---

---
name: slurm-workload-manager
description: Build, configure, run, and test the Slurm Workload Manager from this repository. Trigger when asked to compile Slurm, generate slurm.conf/slurmdbd.conf, start daemons locally, submit or debug jobs, run the test suite, or answer how to use Slurm commands and plugins.
---
# Slurm Workload Manager

Use this skill whenever working with this repo to build or operate Slurm locally for demos, dev, or docs lookups.

## Quick Entry Points
- **Need the right guide?** See `references/doc-map.md` for which doc to open.
- **Need a config quickly?** Start from `references/minimal-slurm.conf.md` and adjust hostnames/paths.
- **Need commands?** `references/cli-cheatsheet.md` lists the common Slurm CLI patterns.

## Prereqs (dev box)
- Linux host with `gcc`, `make`, `python3`, `munge` libs; install dev headers for optional plugins as needed.
- Ensure consistent UID/GID across nodes if testing multi-node (see quickstart_admin.shtml).
- For local single-node demo, you can run all daemons on one host.

## Build From Source (local, single prefix)
1) In repo root: `./configure --prefix=/usr/local --sysconfdir=/etc/slurm` (add plugin enables as needed). If reconfiguring, run `./configure` again; `autoreconf` only if Makefile.am changed.
2) Compile: `make -j$(nproc)`.
3) Optional quick unit tests: `make check` (uses Check; no daemons needed).
4) Install: `sudo make install` (or use DESTDIR for staging).

## Minimal Runtime Setup (single node)
1) Copy `etc/slurm.conf.example` to `/etc/slurm/slurm.conf`; trim partitions/nodes to your host; set `SlurmUser`, `StateSaveLocation`, `SlurmdSpoolDir`, `LogFile` paths that exist and are writable.
2) If using accounting, copy `etc/slurmdbd.conf.example` to `/etc/slurm/slurmdbd.conf`; set MySQL creds and `PidFile/LogFile` dirs.
3) Ensure `munge` is installed and `munged` running; same `munge.key` across nodes.
4) Create needed dirs (`/var/log/slurm`, `/var/spool/slurmctld`, `/var/spool/slurmd`, etc.) and chown to `SlurmUser`.
5) Start daemons (order): `slurmdbd` (if used), `slurmctld`, `slurmd`. For systemd, enable services from generated unit files in `etc/*.service.in` after install.

## Submitting and Inspecting Jobs
- Allocate interactive shell: `salloc -N1 -n1 --time=5:00 bash` then run `srun hostname`.
- Batch job: `sbatch job.sh` where script contains `#!/bin/bash` and `srun ...` lines.
- Status: `sinfo` (nodes/partitions), `squeue` (jobs), `sacct` (history with accounting enabled).
- Cancel: `scancel <jobid>`.

## Debugging
- Logs: see `LogFile` paths from configs; tail `slurmctld.log`, `slurmd.log`, `slurmdbd.log`.
- Check daemon status: `scontrol ping`, `scontrol show daemons`, `scontrol show slurmd`.
- Validate config: `scontrol reconfigure` after edits; `slurmctld -t` for dry-run parse.
- Common issues: wrong `munge` state, mismatched `SlurmctldPort/SlurmdPort`, missing writable dirs, hostname mismatch with `NodeName`.

## Running Test Suites
- Unit tests: `make check` in repo root (no install required).
- Expect tests: install & start daemons, then `cd testsuite/expect` and run `./regression.py` or individual `testX.Y`. Set overrides in `testsuite/expect/globals.local` (e.g., `slurm_dir`, `slurm_conf`, account info).
- Pytest harness: see `testsuite/pytest/README` if present; ensure daemons running and QA deps installed.

## When Editing Slurm Code
- Follow Linux kernel C style (see programmer_guide.shtml#Source-Code).
- After adding files, update relevant `Makefile.am` and run `autoreconf` then `./configure`.
- For plugins, reference `doc/html/plugins.html` for required API hooks and registration macros.

## References
Open only as needed:
- `references/doc-map.md` – which repo docs cover what.
- `references/minimal-slurm.conf.md` – trimmed single-node config skeleton.
- `references/cli-cheatsheet.md` – top commands with examples.