This URL is retained for existing bookmarks. Euler guidance is now split into short canonical pages so access, storage, Slurm, GPU, and tunnel commands cannot drift independently.
Start By Responsibility#
- New CPU user: Euler CPU Track
- New GPU user: CPU track, then Euler GPU Track
- Lab policy: IDEAL Lab Euler Share Policy
Canonical Procedures#
| Need | Guide |
|---|---|
| Access, account activation, and safe SSH keys | Euler Access And SSH |
$HOME, work, scratch, temporary files, permissions |
Euler Storage |
sbatch, squeue, sacct, seff, sstat, arrays, GPUs |
Slurm Job Lifecycle |
| VS Code on an allocated compute node | Euler Tunnel |
| Local ParaView 6.0.0 GUI with Euler processing | ParaView Client/Server |
| Errors and recovery | Euler Troubleshooting |
Practical Labs#
Important Corrections To Older Copies#
- A minimal CPU job does not request 64 CPUs and 250 GB RAM.
- The public share is CPU-only; changing only the account on a GPU job is invalid.
- Current ETH guidance says not to specify a partition unless the workload truly depends on one.
- Current scratch capacity is 2.5 TB per user and files are purged after about two weeks; scratch is not backed up.
- Long or unattended work belongs in
sbatch, not a persistent GPU tunnel.
Primary ETH Documentation#
Verified: 2026-08-21. Review by: 2026-11-21. Owner: IDEAL Lab IT.