annlew / BCsupport

BC support

Notebook Tunneling to interactive compute node on Tetralith

Start an interactive jobLoad a Python module e.g.:

module load Python/3.8.3-anaconda-2020.07-extras-nsc1

Start a jupyter notebook and specify the node you are working on:

jupyter-notebook --no-browser --ip=nXX

example:

[x_alewi@n19 ~]$ jupyter-notebook --no-browser --ip=n19

You will get something like:

[I 11:13:55.182 NotebookApp] The Jupyter Notebook is running at:
[I 11:13:55.182 NotebookApp] http://n19:8888/?token=58a74ec25157356a51e655bca3c47069396d211ebbe2d9e3

To access the notebook, open this file in a browser:

file:///home/x_alewi/.local/share/jupyter/runtime/nbserver-243891-open.html

Or copy and paste one of these URLs:

    http://n19:8888/?token=58a74ec25157356a51e655bca3c47069396d211ebbe2d9e3
 or http://127.0.0.1:8888/?token=58a74ec25157356a51e655bca3c47069396d211ebbe2d9e3

Note the node name (n19) and the port, in this example: 8888

In another terminal, make an ssh tunnel log in, note the node name and port:

ssh -N -L localhost:8888:n19:8888 [e-mail]

Then, in a browser, copy paste the equivalent link:

http://127.0.0.1:8888/?token=58a74ec25157356a51e655bca3c47069396d211ebbe2d9e3

If you only get a link with node name which the browser fails to connect to, try another Python module