Code review comment for lp:~alan-griffiths/mir/cleanup-some-locking

> > I was able to reproduce Jenkins' failure of
> > ServerDisconnect.client_detects_server_shutdown on mako. But only once and
> not
> > repeatably, so can't prove it's due to this branch.
>
> Thanks for trying - I've not been able to reproduce (yet) and your success
> gives me hope.
>
> As it appears consistently on this branch in CI I'm assuming it is either due
> to the branch or exposed by it. Either way I need to track it down to land
> this.

Right, I've been able to reproduce and also on *development-branch && desktop*.

This seems to be a race condition in the client side error reporting - something like detecting the broken pipe on the sending thread "before" processing the disconnect event message on the receiving thread.

I've not yet figured out where it breaks, or why this MP is particularly susceptible on the phone. But look for a separate MP.