Jupyter

There are two ways to use Jupyter notebooks on the KTA Computer System: running your own notebook server on a terminal server, or using the admin-managed public JupyterHub. Use your own server if you need a custom environment or specific packages; use JupyterHub if you just want a quick, ready-to-go notebook without any setup.

Running Your Own Jupyter Server¶

You can launch your own Jupyter server on a terminal server and access it from the browser on your local machine, the same way as the VNC setup.

1. Connect to a terminal server and activate an environment containing Jupyter

$ ssh <LRZ-ID>@nidoking.ktas.ph.tum.de
$ source /opt/anaconda3/bin/activate
$ conda activate MyAwesomeDevEnv

See Anaconda if you don’t have an environment with jupyterlab installed yet.

2. Start Jupyter Lab, specifying a port

$ jupyter lab --no-browser --port=9123

Jupyter will print a URL containing an access token, e.g. http://localhost:9123/lab?token=.... Keep this terminal open, or run it inside tmux or screen so the server keeps running after you disconnect.

3. Forward the port to your local machine

From your local machine, open an SSH tunnel to the same port:

$ ssh <LRZ-ID>@nidoking.ktas.ph.tum.de -L 9123:localhost:9123

4. Open the notebook in your local browser

Paste the URL Jupyter printed in step 2 (including the token) into your local browser. You now have full access to a Jupyter server running on the cluster, in whatever environment you activated in step 1.

Public JupyterHub¶

For quick, interactive use without setting up your own server, each terminal server also runs a shared JupyterHub instance:

• terminal1.ktas.ph.tum.de/jupyterhub

• terminal2.ktas.ph.tum.de/jupyterhub

Log in with your LRZ credentials — the same ones you use for SSH.