Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Launching Jupyter on the FMRIB Cluster

The Jupyter Interactive App allows you to launch a Jupyter session on the FMRIB cluster, which you can then interact with using a web browser.

Before you can launch Jupyter you need to prepare a script that will provide the jupyter command. Typically this would be achieved using a conda environment. As a basic example, create a conda environment `jupyter` using a desktop, shell or SSH session:

conda create -y -n jupyter jupyter

create a text file in your home folder called `jupyter.sh` containing:

#!/bin/bash
. $HOME/.bashrc
conda activate jupyter

In the Jupyter form, enter the full path to this file, /home/fs0/<username>/jupyter.sh in the first text field (first red oval below)

Now choose from Jupyter Notebook, the classic view, or Jupyter Lab, which offers an Integrated Development Environment (IDE) interface.

Specify the startup folder, for example /vols/Scratch/<username>/<myanalysis> in the next field (second red oval).

  • Cores/RAM - choose an appropriate amount of cores and memory for your task. Try to keep these as low as possible, and aim for a maximum of 16GB for every two CPU cores you request (based on one of those cores being a virtual thread).
  • Enter the number of hours you need the session for - this is subject to the maximum time configured on the interactive (or gpu_short) partition.
  • Finally, if your software requires a CUDA device then you can tick the GPU box. This will add two additional fields allowing you to select a specific GPU type and specfiy that you require more than one GPU (default is one GPU). Please be aware that there are a limited number of GPUs you may have to wait some time for your session to start.

Click Launch to request your session.

Open OnDemand JupyterOpen OnDemand Jupyter

 

You will now be taken to the My Interactive Sessions page where you can wait for the job to launch.

Once the card heading turns green (from blue), your session has started and you will find a button 'Connect to Jupyter' that will launch your web session.

NB At present, unlike other interactive apps this will not launch a new tab/window - we do not currently understand why this occurs.