Installation

1. Install Prerequisites: This package requires openconnect which can be installed with:

   sudo apt install -y openconnect

2. Setup Cookie Bookmarklet: Add the following bookmark to your browser's favorites bar (or run the code in your browser console):

   javascript:(function() {  
       var cookieMatch = document.cookie.match(/webvpn=(.*?);/);  
       if (cookieMatch && cookieMatch[1]) {    
           navigator.clipboard.writeText(cookieMatch[1]);    
           alert(cookieMatch[1]);  
       } else {    
           alert(%27No matching cookie found%27);  
       }})();

Usage

1. Simply run sudo ./vpn.sh and follow the on-screen instructions:

2. When you see the Cisco Secure client Download, click your Cookie bookmark or run the Bookmarklet command in the console:

3. Paste the cookie string in the terminal to connect to the VPN.

Usage: ./vpn.sh [option...]

   --kill      Kill the VPN connection
   --help      Display this help message
   --salloc    Specify a custom file for salloc_command
   --server    Server to perform operations on. Default: 'newton'
   --bookmark  Display the bookmarklet code and copy it to clipboard using xclip
Note: This script requires root privileges.

Running a Jupyter Server

1. Login to https://secure.vpn.ucf.edu/
2. Use the bookmarklet from above to get your cookie. It should be automatically copied to your clipboard.
3. Connect to the UCF VPN by running: sudo ./vpn.sh
4. When prompted paste in your cookie from step 2 above.
5. Follow the on-screen instructions to ...
   1. ssh to either newton or stokes (default: newton)
   2. Run the salloc command copied to your clipboard to start an interactive session on the compute cluster
6. Open a new terminal (keep the other one open -- it's interactive!)
7. Run ./startJupyterServer.sh. This starts the Jupyter lab server.

You can now access Jupyter Lab at the URL displayed at the end of the ./startJupyterServer.sh script (look for http://localhost:PORT/lab?token=..., but your PORT will be different).

Custom Anaconda Environments

Once a conda environment is installed as a Jupyter kernel, it's accessible in future Jupyter sessions without needing to specify it during startup. This feature facilitates seamless switching between different environments.

To use your own conda environment from within a Jupyter notebook, first install the ipykernel library. Since package installation typically requires multiple threads (which restricts your ability to do so on the user side of the cluster), you should do this from the compute side of the cluster as follows:

1. Connect to the user side of the cluster with either ssh newton.ist.ucf.edu or ssh stokes.ist.ucf.edu
2. Start an interactive session on a compute node with srun --time=0:30:00 /bin/bash
3. Create or activate the conda environment with conda activate $ENV_NAME
4. Run conda install -c anaconda ipykernel pylint to manage kernels and ensure linting works correctly
5. Install any other packages/dependencies you need
6. Add your conda environment to the jupyter server kernels list with:

   python -m ipykernel install --user --name "$ENV_NAME" --display-name "Python ($ENV_NAME)"

7. Close the connection on the compute side with exit.

The conda environment is now installed as a Jupyter kernel. You can access it in the browser or in VS Code (see below).

VS Code integration

You can use the compute cluster to run your Jupyter Notebook directly within VS Code. Follow these steps:

1. Open VS code
2. Open any ipynb file
3. From the command palette (Ctrl+Shift+P), select Notebook: Select Notebook Kernel > Existing Jupyter Server > Paste the localhost url (look for http://localhost:PORT/lab?token=... from the output of ./startJupyterServer.sh. Your PORT will be different)

Notes:

• By default, your token (also called password in VS Code) is your newton username. Just type whoami while connected to newton. You can change this by modifying the script.
• Adding a kernel for the first time can take a few moments (~30s); monitor the status from the startJupterServer.sh terminal. If it seems to be taking too long or you see a 403 Forbidden error, just click Interrupt and try adding the kernel again

Acknowledgement

Thanks to Dr. R. Paul Wiegand who wrote the original scripts (available in the archive folder). Special thanks to Sonny Bhatia who wrote vpn.sh, the original bookmarklet, and pivoted the original scripts to their interactive versions.