Cluster Access
In this page you will find information about how to connect to SCI-UC clusters. You can do it via SSH (CLI) using public key method.
SSH Key Requirement
Access to the cluster is only allowed via SSH key-based authentication. Password logins are disabled.
If you already have an SSH key pair, simply attach your public key file to your request email.
If you need to generate a new SSH key pair, run the following command on your local computer:
ssh-keygen -t ed25519
You can press Enter to accept the default file location (~/.ssh/id_ed25519). Once generated, attach the public key file (id_ed25519.pub) to your access request.
Once you have access to the cluster, you can allow any aditional SSH public key to authenticate on the User Interface node by following these instructions.
Access and Use of the Cluster
Once your request is approved, you’ll receive your assigned username and access confirmation.
You can log in to the cluster’s user interface node (front-end) using SSH:
[user@localmachine]$ ssh user@ui.sci.unican.es
Access from Windows
If you are using Windows, you can connect using an SSH client such as PuTTY. In this documentation page you can find the instructions for accesing the cluster using PuTTY.
Once you have succesfully logged in
Once connected to the front-end node, you can:
- Submit jobs to the compute nodes using the cluster’s scheduler (slurm).
- Access shared storage areas (e.g., $HOME, lustre).
- Transfer files using scp or rsync.
UI Node Resource Usage Limits
To ensure fair usage and system stability, the following resource limits apply to all users on the login (UI) node.
You can inspect your current limits using the ulimit command:
ulimit -a
Enforced Limits
| Resource | Soft Limit | Hard Limit | Description |
|---|---|---|---|
| Memory (virtual) | 2 GB | 4 GB | Total virtual memory per process |
| Processes (nproc) | 512 | 1024 | Maximum number of processes or threads |
| CPU time | 120 minutes | 240 minutes | Max CPU time per process (accumulated) |
- Soft limits can be increased by the user, up to the hard limit.
- Hard limits are the maximum values enforced by the system.
For heavy computations or long-running tasks, use the Slurm batch system instead of the login node.
Cron Jobs on the UI Node
Users are allowed to schedule background jobs using the crond service on the login node.
Editing your personal cron jobs:
crontab -e
Viewing your scheduled jobs:
crontab -l
Note: Cron jobs are subject to the same resource limits as interactive sessions. Avoid running heavy jobs via cron on the UI node.
Tips
- Use
srun,sbatch, orsallocfor compute jobs. - Use cron only for lightweight automation (e.g., syncing files, sending notifications, or monitoring job status).