Shutting kernel and jupyterlab doesn't kill MATLAB or proxy process

[asfilion]

I can use the proxy with R2023a and Python 3.10 on Windows 11. When I shut down the MATLAB kernel and shut down jupyterlab, the MATLAB process and matlab-proxy-app.exe process are still active. If I reopen jupyterlab and use the proxy again, I'll get another MATLAB and proxy process. I don't see any way to shut these processes other than manually killing them in the task manager.

[prabhakk-mw]

Thank you @asfilion, I've been able to reproduce this issue and we are working towards a fix.

[rashedmyt]

Hi @asfilion

MATLAB Integration for Jupyter currently uses jupyter-server-proxy to provide both MATLAB in browser and MATLAB Kernel functionalities. Unfortunately, jupyter-server-proxy doesn't plug-in to the shutdown infrastructure provided by jupyter.

However, if you have access to the terminal from which JupyterLab is launched, you can use CTRL+C to shutdown both JupyterLab and MATLAB Integration.

There is a pending PR on jupyter-server-proxy repository which provides more access to manage the integration. For more information, please look at jupyterhub/jupyter-server-proxy#395

Hope this helps.

[krisctl]

This restriction has been now lifted and one can now shutdown matlab-proxy and corresponding MATLAB via the MATLAB Kernel shutdown* and JupyterLab Shut Down from the JupyterLab interface. Please see the attached video for demo purposes.

* When multiple MATLAB Kernels are accessing the same backend matlab-proxy, termination of matlab-proxy is delayed until the last reference to the backend matlab-proxy is not cleaned up

JMP.55.mp4