Overview
Login nodes are shared gateways to the clusters, designed for light tasks: editing files, submitting jobs, compiling code, and monitoring job progress. Heavy computation should always run on compute nodes.
Appropriate use
- Editing scripts, viewing files, and managing data
- Compiling code and loading modules
- Submitting and monitoring SLURM jobs (
sbatch,squeue) - Lightweight testing or small-scale calculations
What to avoid
- Running long or resource-intensive computations directly on login nodes
- Starting multi-core or large-memory workloads on login nodes
- Leaving idle processes consuming memory or CPU
Tips
- Log out when finished
- Use
srun --pty bashor batch scripts for interactive or large jobs on compute nodes - Monitor your login-node usage and respect CPU/memory limits
- Contact support via iacs.supportsystem.com for job-placement help
Key message: Login nodes are a shared resource. Following these guidelines ensures all users have a stable and responsive environment.
Applies to