Managing Lifecycle of nvdashboard

rapidsai / jupyterlab-nvdashboard

A JupyterLab extension for displaying dashboards of GPU usage.

BSD 3-Clause "New" or "Revised" License

583 stars 77 forks source link

Managing Lifecycle of nvdashboard #63

Open aroraakshit opened 4 years ago

aroraakshit commented 4 years ago

Currently there is no way to close an existing nvdashboard process. I have tried multiple ways to close the process but it starts back up again automatically. I believe it is because of ioloop module by Tornado web server.

Please include a safe way to:

Shut it down
Restart it with a custom port

This will help manage nvdashboard server lifecycle in an independent way (as opposed to being tied to a default port).

jacobtomlinson commented 4 years ago

Thanks for raising this. Could you share how you are starting the extension? Via Jupyter Lab or standalone?

aroraakshit commented 4 years ago

The extension is automatically started as part of RAPIDS NGC Container. So I assume within the container it is initiated with the jupyter lab.

jacobtomlinson commented 4 years ago

Ok thanks. Do you have thoughts on what the most natural feeling way to restart and reconfigure would be?

aroraakshit commented 4 years ago

From a user experience point of view, a terminal command to restart nvdashboard (and an argument that specifies the new port) that can be executed inside the container where jupyter lab is running, would be great.