[dashboard] Gracefully handle connection disposal

[shaal]

Bug description

Loading a workspace took very long time, at the end I got this:

Steps to reproduce

It happened only once; But I've never seen it before.
I was able to open the URL in a new tab and it worked as expected.

This is the URL:
https://gitpod.io/#DP_PROJECT_NAME=feeds_ex,DP_ISSUE_FORK=feeds_ex-3025848,DP_ISSUE_BRANCH=3025848-context-examples-with-code,DP_PROJECT_TYPE=project_module,DP_MODULE_VERSION=8.x-1.x,DP_CORE_VERSION=9.2.0,DP_PATCH_FILE=,DP_INSTALL_PROFILE=standard/https://github.com/shaal/drupalpod

Expected behavior

It would be awesome to know what is going on while it's happening, why a workspace takes long, what part of it got stuck, is it something I did wrong, or a glitch in the system. I'm just happy it only happened once :)

Example repository

https://gitpod.io/#DP_PROJECT_NAME=feeds_ex,DP_ISSUE_FORK=feeds_ex-3025848,DP_ISSUE_BRANCH=3025848-context-examples-with-code,DP_PROJECT_TYPE=project_module,DP_MODULE_VERSION=8.x-1.x,DP_CORE_VERSION=9.2.0,DP_PATCH_FILE=,DP_INSTALL_PROFILE=standard/https://github.com/shaal/drupalpod

Anything else?

No response

[jankeromnes]

Hi @shaal! Thanks for raising this issue. 😁

I believe that "Connection got disposed" is an error that typically gets logged by the front-end when it cannot reach the back-end at all (e.g. when there is an Internet outage, or even a brief network flake out).

I was able to reproduce it by loading the dashboard, then turning off my laptop wifi, and then trying to navigate to a different page & back.

[akosyakov]

Probably dashboard should handle reconnections more gracefully. As far as I know the global load balancer drops web sockets each 10 mins, so it does not even need to be an issue on the client side.

[akosyakov]

I got the same today as well.

[csweichel]

/schedule

[rl-gitpod]

/unassign

[AlexTugarev]

Looking at the code this is not a websocket connection issue at all.

That error is propagated from server to the frontend.