fasrc-ood-jupyter

OnDemand app for serving Jupyter notebooks out of your local path, so you can conveniently manage Jupyter extensions according to your preferences.

News: This repository has been updated to ensure compatibility with the FAS RC Rocky upgrade in June 2023.

Prerequisites

You must have a conda/mamba environment with jupyter lab/notebook installed inside. According to the FASRC documentation, here is a step by step guide about how to set one:

1. Connect to the RC VPN and log in to the RC cluster via SSH:

   ssh user_name@login.rc.fas.harvard.edu

2. In your ssh session, Load the Python module, which will automatically load the Mamba module:

   module load python/3.10.9-fasrc01

   Confirm that the modules are loaded by running module list:

   Currently Loaded Modules:
   1) Mambaforge/22.11.1-fasrc01   2) python/3.10.9-fasrc01

3. Configure your .condarc file to install environments in the lab tier 1 storage for better performance:

   mkdir /n/hekstra_lab/people/<user_name>/envs

   Add the following block to your ~/.condarc file. Create the file if it doesn't exist:

   envs_dirs:
      - /n/hekstra_lab/people/<user_name>/envs

4. Create a Mamba environment with a custom name:

   mamba create -n <env_name> python=3.10

   Note: For GPU-accelerated deep learning frameworks, refer to the documentation on setting up compatible environments: pytorch, tensorflow

5. Activate the newly created environment, install Jupyter Lab (required), and any additional packages you need:

   mamba activate <env_name>
   pip install jupyterlab
   pip/mamba/conda install <other_packages_you_need>

Installation

1. First, you need to enable the FAS RC VPN
2. Create a new folder in your home directory on the RC cluster mkdir ~/.fasrcood/dev
3. Navigate to the Open OnDemand app page
4. Click on the </> Develop dropdown (top right) and go to "My SandBox Apps (Development)"
5. Click on "New App" (top left)
6. Click on "Clone Existing App"
7. Fill in some suitable directory name. This will determine the name of the directory inside ~/fasrc/dev used for this project
8. Paste the following into the "Git remote" dialog box

   https://github.com/Hekstra-Lab/fasrc-ood-jupyter.git

9. Click "Submit"

Launching

1. Go to the vdi interactive session
2. Click on "CustomizedJupyter" (left bottom, Interactive Apps Sandbox section)
3. Fill out the form and click submit

When the notebook is ready, a "Connect to Jupyter" button will appear.

RC VPN

• General instructions for the RC VPN are available in the FASRC docs
• Linux specific instructions using openconnect here