Step 1 — Get an Account

Access to pasteur requires an account provisioned by a cluster administrator. Contact Dr. Garibay or the CASL lab admins to request access. Provide your preferred username and university affiliation.

Step 2 — Connect via SSH

All user access is through the login node pasteur-login. You do not SSH directly to the compute node.

$ ssh your-username@10.196.20.50
Welcome to pasteur-login (Ubuntu 24.04 LTS)
CASL HPC Cluster — Login Node
Do not run compute jobs on this node. Submit jobs via Slurm.bash

Replace your-username with your cluster account name.

SSH Key Setup (Recommended)

Password authentication works but key-based authentication is faster and more secure. Generate a key pair on your local machine if you don't already have one:

# On your LOCAL machine
$ ssh-keygen -t ed25519 -C "your-email@university.edu"
$ ssh-copy-id your-username@10.196.20.50local

After this, SSH will authenticate automatically using your key.

Windows Users (MobaXterm)

If you're using MobaXterm:

1. Create a new SSH session with the login node hostname
2. Set your username in the session settings
3. Use MobaXterm's built-in key manager to load your private key (optional but recommended)
4. Save the session for future use

Step 3 — Explore Your Environment

Once logged in, your home directory is on the shared /data filesystem — the same paths are visible on both pasteur-login and the compute node:

$ pwd
/data/home/your-username

$ df -h /data/home
Filesystem      Size  Used Avail Use% Mounted on
pasteur:/data/home   14T  ...  ...  ...  /data/homepasteur-login

Your storage spans several areas — the exact paths include your group name, which is assigned when your account is created. See the Storage Layout page for details:

• /data/home/your-username — personal home directory
• /data/scratch/<group>/your-username — fast temporary workspace (not backed up)
• /data/apps/<group>/your-username — conda environments and personal software
• /data/shared/ — shared project data and archives

Step 4 — Verify Slurm Access

Check that you can see the Slurm cluster and your account:

# Check cluster status
$ sinfo
PARTITION  AVAIL  TIMELIMIT  NODES  STATE  NODELIST
debug         up   1:00:00      1   idle   pasteur
standard*     up  2-00:00:00   1   idle   pasteur
extended      up  7-00:00:00   1   idle   pasteur

# Check your account limits
$ sacctmgr show assoc user=$(whoami) format=user,partition,qos,maxgres -p
User|Partition|QOS|MaxGRES|
your-username||normal||pasteur-login

Step 5 — Run a Test Job

Submit a simple test job to verify your setup is working end-to-end:

# Submit an interactive test job
$ srun --partition=debug --time=00:05:00 --pty bash

# Inside the job — you are now on the compute node
$ hostname
pasteur

$ echo "Hello from the compute node!"
Hello from the compute node!

$ exit   # exit the interactive sessionpasteur-login

Or submit a batch job:

# Create a simple test script
$ cat > test_job.sh << 'EOF'
#!/bin/bash
#SBATCH --job-name=test
#SBATCH --partition=debug
#SBATCH --time=00:02:00
#SBATCH --output=test_%j.out

echo "Job ID: $SLURM_JOB_ID"
echo "Running on: $(hostname)"
echo "Date: $(date)"
EOF

# Submit it
$ sbatch test_job.sh
Submitted batch job 42

# Check its status
$ squeue -u $(whoami)
  JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST
     42     debug     test username  R       0:03      1 pasteur

# View the output once it completes
$ cat test_42.out
Job ID: 42
Running on: pasteur
Date: Thu Mar 26 12:34:56 UTC 2026pasteur-login

Next Steps