Slurm Node Dashboard

A Next.js dashboard for operating Slurm clusters, with a configurable AI assistant built for HPC workflows.

Slurm Node Dashboard combines a live node view with optional historical analytics, GPU reporting, software modules, and a configurable Slurm assistant that can answer questions, call cluster tools, and be managed from the admin UI.

Highlights

• Live cluster view with CPU/GPU node status, filtering, rack/group layouts, and 15 second refresh polling
• Optional feature-flagged pages for software modules, cluster rewind, and job metrics
• GPU analytics from Prometheus/DCGM recording rules, including job-level utilization views
• AI assistant with streaming chat, built-in Slurm workflow tools, documentation search, and structured tool result cards
• Admin-managed assistant configuration with YAML overrides, custom tools, restricted topics, and prompt preview/testing
• Admin dashboard for partitions, reservations, QoS, plugins, hierarchy data, and LLM assistant settings

Requirements

• Node.js 20.9.0 or newer
• npm
• Access to a Slurm REST API (slurmrestd)
• Optional: PostgreSQL for historical metrics, rewind, and hierarchy features
• Optional: Prometheus/DCGM metrics for GPU utilization features
• Optional: OpenAI-compatible API credentials for the assistant

Node 20 LTS and Node 22 both satisfy the current engine requirement.

Quick Start

git clone https://github.com/thediymaker/slurm-node-dashboard.git
cd slurm-node-dashboard
npm install

Create a .env.local with the minimum local settings:

NEXT_PUBLIC_BASE_URL=http://localhost:3020
BETTER_AUTH_SECRET=replace-with-a-long-random-string
ADMIN_USERNAME=admin
ADMIN_PASSWORD=change-me

CLUSTER_NAME=My Cluster
SLURM_SERVER=slurmrestd.example.edu
SLURM_SERVER_PORT=6820
SLURM_PROTOCOL=http
# Match the version exposed by your slurmrestd, for example v0.0.40
SLURM_API_VERSION=v0.0.40
SLURM_API_ACCOUNT=dashboard
SLURM_API_TOKEN=replace-with-your-token

Then start the app:

npm run dev

Open http://localhost:3020.

Documentation

For full deployment, feature flag, AI assistant, GPU metrics, and historical data setup details, see slurmdash.com.

The optional historical ingestor is documented in slurm-history-ingestor/README.md and slurm-history-ingestor/DOCUMENTATION.md.

Contributing

Pull requests are welcome. Keep changes focused, run linting for touched areas, and include any relevant setup or behavior notes in the PR.

License

This project is licensed under the GNU General Public License v3.0. See LICENSE.md for details.

Support

• Documentation: slurmdash.com
• Issues: GitHub Issues
• Contact: Johnathan Lee - john.lee@thediymaker.com

---

Made with ❤️ for HPC