DNS failure and delays break connections

[pacien]

In case of DNS resolver malfunction (unavailability or slow response),
the delay introduced by DNS queries causes other connections to get closed.

In the worst case, this causes a complete disconnection of the node to its tinc
network, without being able to reconnect (looping with the same issue).

The root issue is that name resolution queries seem to be done synchronously?

A workaround for this is to only use IP addresses in Address fields,
but this is not always possible (for example for hosts with dynamic addresses
relying on dyndns).

tincd: Ready
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Error looking up [REDACTED] port 655: Temporary failure in name resolution
tincd: Awaking from dead after 90 seconds of sleep
tincd: Forcing connection close after sleep time [REDACTED] ([REDACTED] port 655)
tincd: Forcing connection close after sleep time [REDACTED] ([REDACTED] port 655)
tincd: Forcing connection close after sleep time [REDACTED] ([REDACTED] port 655)

Tinc version 1.1pre18

[splitice]

I have a WIP pr for this.

Stuck on failing ci but I can't view the ci logs to find out why.