Retry a few times when the initial connection fails

[sunu]

Sometimes the initial websocket connection fails. This is a workaround that tries to connect a few times with some delay in between connection attempts when a request fails.

Relates to #105. Although it doesn't solve the actual underlying issue, I think it's a reasonable fix from a UX point of view.

Here's a video of the patch in action:

desktop-proxy-retry.mp4

[welcome]

Thanks for submitting your first pull request! You are awesome! 🤗

If you haven't done so already, check out Jupyter's Code of Conduct.

You can meet the other Jovyans by joining our Discourse forum. There is also a intro thread there where you can stop by and say Hi! 👋

Welcome to the Jupyter community! 🎉

[github-actions]

👈 Launch a binder notebook on this branch for commit 975d084

I will automatically update this comment whenever this PR is modified

[yuvipanda]

To increase review capacity, I'm going to ask @batpad - have you got a moment to give this a quick review? Happy to merge if it looks good to you

[sunu]

Some additional notes:

I was trying to use this fix with https://github.com/sunu/jupyter-remote-qgis-proxy which starts a QGIS instance inside a tornado request handler. I noticed that that although the VNC connection is successfully established through the retry, the started QGIS instance remained in the background not attached to the VNC session.

I had to add an additional workaround in the tornado handler that checks if websockify is already running and if not starts the QGIS process with a delay after waiting for websockify to start: https://github.com/sunu/jupyter-remote-qgis-proxy/blob/bab4ce8c66ef6b3b377f32a8827f8335d75c6816/jupyter_remote_qgis_proxy/handlers.py#L17

[yuvipanda]

So, I've lots of opinions on how to make this code better, similar to what I tried in #81.

However, I'm also trying to refocus on faster review and merges instead, so we become a more welcoming and vibrant project. With that in mind, my approach turns to - 'does this work? Is it better than what we have now? does it move us significantly in the wrong direction?'

The answer to all these is 'yes, yes, no' so I'm going to merge this one.

Thanks a lot @sunu!

[yuvipanda]

@sunu jupyter-remote-desktop-proxy 2.0.1 has been released with this patch in it! Welcome to your first released contribution to the jupyterhub ecosystem (I think?). Hope to see more!

[sunu]

amazing, thank you @yuvipanda!