Workspace teardown is unreliable

[csweichel]

Bug description

There seems to be a timing issue where ws-daemon stops the IWS before workspacekit had a change to tear down the workspace. This causes workspaces to be stuck in "stopping".

{"exitCode":0,"level":"debug","message":"supervisor exit","serviceContext":{"service":"supervisor","version":""},"severity":"DEBUG","time":"2021-07-26T08:22:31Z"}
{"@type":"type.googleapis.com/google.devtools.clouderrorreporting.v1beta1.ReportedErrorEvent","error":"rpc error: code = Unavailable desc = connection error: desc = \"transport: Error while dialing dial unix /.workspace/daemon.sock: connect: connection refused\"","level":"error","message":"cannot trigger teardown","ring":0,"serviceContext":{"service":"workspacekit","version":""},"severity":"ERROR","time":"2021-07-26T08:22:31Z"}

Steps to reproduce

Start and stop enough workspaces in core-dev.

Expected behavior

No response

Example repository

No response

Anything else?

No response

[csweichel]

/schedule

[csweichel]

probably caused by #5079 - closing this one for now. if we find that was not the cause we'll re-open.