Notes
- Job form. Researchers submit Slurm jobs without a terminal: script, resources, inputs.
- Queue. Status and logs per job, fed by a Node service that talks to Slurm for the user.
- ERDA picker. Inputs come from, and results go back to, AU's own ERDA storage.
- Nodes. Compute nodes on Oracle Cloud scale up with the queue and down to zero when it's empty.
- Login. OAuth with Entra ID: university accounts only.
what & whyA scalable, easy-to-use compute job system for researchers: Slurm's power behind a React/Node web app, with both web stack and compute scaling on Oracle Cloud.
interestingScaling the web tier and the compute pool separately, since a busy week of uploads isn't a busy week of jobs.
| REV | DATE | CHANGE |
|---|---|---|
| A | 2025 | first release, one group |
| B | 2026 | autoscaling compute |