I’m going to assume you already have a conda environment setup. If you don’t refer to a the previous CampusCluster post.
To install jupyter notebook/lab do:
> conda install -c conda-forge jupyterlab
Next we need to setup a remote notebook. In short, first you got to get a jupyter server running on the remote machine. You also got to specify the port it is attached beforehand so you can connect to it from your local machine. Use this command:
> jupyter notebook --no-browser --port=8889
Now you should see some startup text, but you should also see a link like:
To access the notebook, open this file in a browser:
file:///home/kani/.local/share/jupyter/runtime/nbserver-14164-open.html
Or copy and paste one of these URLs:
http://localhost:8889/?token=b2c5e1eae3afa5d217e19af83cc742e1cdce7fb7784451fb
or http://127.0.0.1:8889/?token=b2c5e1eae3afa5d217e19af83cc742e1cdce7fb7784451fbThe token string in the url (last line) is the important thing.
Next on the local machine, we got to forward one of our ports to the port on the server. we can do this using the ssh -L:
> ssh -N -f -L localhost:8888:localhost:8889 netid@cc-login.campuscluster.illinois.edu
There won’t be an output to this command if it works. You’ll be unsure if it did anything but if you try to run it again you’ll see a message that port 8888 is in use.
You can also check to see the ports that are in use by using the following command:
> lsof -i -P -n | grep LISTEN
The command will setup a ssh instance that forwards everything going to port 8888 on your local machine to port 8889 on the campuscluster which is attached to your still running jupyter server.
Next we got to interact with our jupyter server. To do that, open a webbrowser and type:
localhost:8888
If you did everything right, this should open up a jupyter page asking for a password and/or token. We didn’t setup a password but we do have the token from when we setup the server. Enter that token and you should be greeted by the familiar jupyter file browser where you can do whatever you want.
To kill your session you can use CTRL-C to kill the server, and on your local machine you got to kill the port forwarding. To do this you can use:
> lsof -n -i4TCP:8888
To find the PID of the ssh process using the port and then do:
> kill PID
To kill it.
References
- Instructions I adapted for this post - https://amber-md.github.io/pytraj/latest/tutorials/remote_jupyter_notebook