Jupyter notebooks are accessed via your web browser. This makes - in principle - rather easy to run notebooks on the CIP hosts and access them from your own machine at home, but there are some caveats to observe. You should access the CIP cluster via the (virtual) SSH gateway ssh.cip.ph.nat.tum.de, but you should run your jupyter server on one of the more powerful physical machines. You need to tunnel the communication with the jupyter server through a two-stage SSH connection. (Additionally, an eduVPN connection is required to access the TUM computers from outside the MWN.) From your own machine (e.g. at home), type
ssh gi12abc@cip2ryzen1.cip.ph.tum.de -J gi12abc@ssh.cip.ph.nat.tum.de -L 8888:cip2ryzen1:8888
where obviously gi12abc is your own login. You can use any of the CIP machines, but we recommend cip2ryzen1-5. Please observe that the hostname that you connect to needs to match the one in the port redirect at the end of the command! (We need to use the hostname rather than 127.0.0.1 because otherwise the jupyter token authentication does not work.)
Once you are logged into the CIP host, you can start the jupyter server, but you might first want to enter the virtual python environment that we provide for machine learning projects,
source /mount/packs/ml/bin/activate
jupyter-notebook-remote
In the output of the last command, find the line starting with "http://127.0.0.1:8888/tree?token=3d7d8..." and copy this to your local browser. You should now be connected to the jupyter notebook server on the CIP computer.
If your output contains the line "The port 8888 is already in use, trying another port.", check which port was actually assigned to you, then stop the jupyter-notebook-server with CTRL-C, logout and do the whole procedure again, replacing every 8888 with the new port number.