Running a Long-Running Jupyter Notebook Remotely using tmux

jupyter_tmux.md

Running a Long-Running Jupyter Notebook Remotely using tmux

Step-by-Step Instructions:

1. Start tmux Session

   • Connect to your remote server via SSH:

     ssh user@remote-server

   • Start a tmux session to keep your terminal session active even after disconnection:

     tmux

   • Optionally, you can name the session for easier access later:

     tmux new -s my_jupyter_session

2. Launch Jupyter Notebook

   • In the tmux session, navigate to the folder containing your Jupyter notebooks and start the Jupyter server:

     jupyter notebook --no-browser --port=8888

   • Note down the URL with the token that appears in the terminal output. You will need this to access the notebook later.

3. Detach from the tmux Session

   • To leave the tmux session without stopping it, press:

     Ctrl+b, then d
     

   • This will detach the tmux session, keeping your Jupyter server running.

4. Access the Jupyter Notebook from Your Local Machine

   • Create an SSH tunnel to forward the Jupyter port to your local machine:

     ssh -L 8888:localhost:8888 user@remote-server

   • Open your browser and go to:

     http://localhost:8888
     

   • Use the token noted earlier to access your running Jupyter notebook.

5. Reattach to tmux Session (Optional)

   • If you need to reattach to your tmux session to check on Jupyter:

     tmux attach -t my_jupyter_session

Running a Long-Running Jupyter Notebook using VSCode Jupyter Extension

If you are using VSCode's Jupyter extension and want to keep the kernel running even after disconnecting from the server, follow these steps:

1. Start a Jupyter Server in tmux

   • SSH into your remote server and start a tmux session:

     ssh user@remote-server
     tmux new -s my_jupyter_session

   • Start the Jupyter server in the tmux session:

     jupyter notebook --no-browser --port=8888

   • Note down the URL and token displayed.

2. Configure VSCode to Use the Running Jupyter Server

   • On your local machine, create an SSH tunnel to forward the Jupyter port:

     ssh -L 8888:localhost:8888 user@remote-server

   • Open VSCode and press Ctrl+Shift+P (or Cmd+Shift+P on macOS) to open the Command Palette.
   • Select "Jupyter: Specify Jupyter Server for Connections" and enter the URL of your running Jupyter server, such as:

     http://localhost:8888
     

   • Use the token you noted earlier to connect.

3. Detach from tmux

   • Detach from the tmux session to keep the Jupyter server running in the background:

     Ctrl+b, then d
     

Checking if Jupyter is Running Without Reconnecting

To check if your Jupyter server is still running without reattaching to tmux, you can use the following commands:

1. List Running Jupyter Notebook Servers

   jupyter notebook list

   This command will display all currently running Jupyter servers, including their URLs and tokens, if they are still active.

   Example output:

   Currently running servers:
   http://localhost:8888/?token=TOKEN_HERE :: /path/to/notebooks
   

2. List Running Jupyter Processes

   • Use the following command to check all Jupyter processes by your user:

     ps aux | grep jupyter | grep $(whoami)

   • This will list all Jupyter-related processes for the current user $(whoami). Look for entries containing jupyter-notebook or kernel to confirm they are still running.

   Example output:

   melashri  12345  0.5  1.0 456789 123456 ?  Sl   14:30   0:20 /usr/bin/python3 /path/to/jupyter/runtime/kernel-XXXX.json
   

Summary of what to do

• Use tmux to start and keep a Jupyter server running on a remote server.
• Detach from the tmux session to leave it running in the background.
• Use SSH tunneling to access the Jupyter notebook from your local machine.
• Configure VSCode to connect to the running Jupyter server if using the VSCode Jupyter extension.
• To check if the Jupyter server is still running, use jupyter notebook list or ps aux | grep jupyter.