Remote Jupyter Notebook on shell servers

See also

We now have a General use student/teaching JupyterHub installation which may serve your uses more simply.

Here we describe how you can utilise Aalto computing resources for Jupyter Notebook remotely. The guide is targeted for UNIX users at the moment.

Aalto provides two “light computing” servers:, We demonstrate how to launch a Jupyter Notebook on brute and access it on your laptop.

alternate text

< System activity on Brute >


# Create your Kerberos ticket

# Create a session. I use tmux

# Load the Aalto Scicomp python environment
module load scicomp-python-env

# Create your env
conda create -n env-name python=3.6 jupyter

# Activate your python environment
source activate env-name

# Launch jupyter notebook in headless mode and a random port number
jupyter notebook --no-browser --port=12520


You might get messages like The port 12520 is already in use, trying another port while starting the notebook server. In that case, take note of the port the server is running in, e.g.:

[I 15:42:14.187 NotebookApp] The Jupyter Notebook is running at:
[I 15:42:14.187 NotebookApp] http://localhost:12470/?token=kjsahd21n9...

and replace “12520” below with the correct port number, 12470 in this case.

Now back to your laptop

# Forward the port
ssh -L 12520:localhost:12520 -N -f -l username

Now launch your browser and go to http://localhost:12520 with your token.