Quickstart

Modified

March 19, 2026

Abstract

This page provides a condensed checklist to get you from zero to your first running job. Each step links to the full documentation – please take the time to read through the complete User Guide as you go.

This page is a shortcut, not a substitute. The sections linked below contain important details on security, storage, and job configuration that are essential for productive use of the cluster.

  1. Get a GSI Linux Account – Request an account through the GSI account registration. Processing may take multiple days, so plan ahead.

  2. Request a Slurm Account Association – Contact your account coordinator or fill out a user access request. If no coordinator exists for your group, specify default as the requested account.

  3. Set Up SSH Key Authentication – The cluster requires key-based login. Generate a key pair and install it on the login nodes. See Key Authentication.

  4. Configure SSH Proxy Jump – If connecting from outside GSI, set up a proxy jump through lxlogin.gsi.de. See Proxy Jump.

  5. Log In to a Submit Node – Connect to a VAE submit node, e.g. ssh vae25.hpc.gsi.de. See Submit Nodes for the full list.

  6. Identify Your Lustre Working Directory – Your home directory (/u/$USER) is not available on compute nodes. All job data must reside on Lustre shared storage. Ask your account coordinator for the path. The typical convention is:

    export LUSTRE_HOME=/lustre/$(id -g -n)/$USER

  7. Submit Your First Job – Create a test script and submit it:

    cat > $LUSTRE_HOME/hello.sh <<'EOF'
#!/bin/bash
hostname ; date ; sleep 30
EOF
chmod +x $LUSTRE_HOME/hello.sh
sbatch --chdir $LUSTRE_HOME -- $LUSTRE_HOME/hello.sh
squeue --me

    For a detailed walkthrough including job monitoring, error handling, and wrapper scripts see First Job.