Closed hannesm closed 11 months ago
I think the entire dns_server_mirage and dns_mirage logic may need an overhaul with a view on when which failures may arise, and how to properly handle them (to avoid such tight loops). maybe timer
is the only thing, though...
I think this improves the situation so I will merge. If there are unintended effects we can revisit.
If there's a (Dns_server.Secondary.)timer-triggered connection request, do not infinitely try to connect (via request -> fail -> close -> request) in a tight loop, but cancel on first sight (the timer is responsible for re-connecting)
I'm in the process of testing this in production.