Jupyter Notebook

ExCl → User Documentation → Jupyter Quick Start

Installing Jupyter

Since there are many ways to install Jupyter using various python management tools, I will not reproduce the documentation here. The official documentation for installing Jupyter can be found at Project Jupyter | Installing Jupyter. However, I will highlight the methods of using Jupyter with UV, running Jupyter Notebooks in VS Code, and the alternative to Jupyter notebooks, Marimo | ExCL User Docs. These methods are all the methods that I typically use when working with python notebooks.

Jupyter with UV

See the UV documentation, . This documentation is well written and covers:

Jupyter kernels using virtual environments

See . Although , the following steps are still a good way to manually create and use a kernel from Jupyter.

Create a python virtual environment and activate it. Then install ipykernel and then install the kernel for use in Jupyter.

Use jupyter kernelspec list to view all the installed Jupyter kernels.

To uninstall a Jupyter kernel use uninstall.

Accessing a Jupyter Notebook Running on ExCL

A Jupyter notebook server running on ExCL can be accessed via a local web browser through port forwarding the Jupyter notebook's port. By default, this is port 8888 (or the next available port). This port might be in use if someone else is using running a notebook. You can specify the port with the --port flag when launching the Jupyter notebook. To use a different port just replace 8888 with the desired port number. In order to port forward from an internal node, you have to port forward twice, once from your machine to login.excl.ornl.gov and once again from the login node to the internal node (i.e. pcie).

Using FoxyProxy (Recommended)

You can use directly connect to the server running on a worker node. If I’m running a Jupyter Notebook on zenith2 with the command jupyter notebook, I can connect to the notebook directly using the printed message:

I hold control on Windows/Linux and click the URL to open it in my local browser.

Detailed instructions for Linux/Mac

These instructions go over how to access a Jupyter notebook running on the pcie node in the ExCL Cluster. If you want to access a different system, then replace pcie with the system you intend to access.

1. Specify the ports that you intend to use. Choose a different number from the default so that you don't conflict with other users.

2. From your local machine connect to pcie using login.excl.ornl.gov as a proxy and local forward the jupyter port.

3. (Optional) Load the anaconda module if you don't have jupyter notebook installed locally.

If you ssh client is too old for proxyjump to work, you can always break up the process into another step.

1. From your local machine connect to login.excl.ornl.gov and local port forward port 8888.

2. From the login node connect to pcie and local port forward port 8888

3. Launch the Jupyter server on pcie

Detailed instructions for Windows with MobaXterm

These instructions go over how to access a Jupyter notebook running on the pcie node in the ExCL Cluster.

1. From your local machine connect to login.excl.ornl.gov using MobaXterm.
2. Go to tools and click on MobaSSHTunnel. Use MobaSSHTunnel local forward port 8888.

   Click on MobaSSHTunnel

Detailed instructions for Windows with Visual Studio Code

These instructions go over how to access a Jupyter notebook running on the quad00 node in the ExCL Cluster using Visual Studio Code to handle port forwarding.

1. Open Visual Studio Code

2. Make sure you have the Remote - SSH extension installed.
3. Setup .ssh