Post #2358488
2026-05-02 07:39 UTC
@nygren@hachyderm.io @phils@chaos.social Interesting... Would this be worse than when an IPv4-only recursive resolver encounters an IPv6-only authoritative nameserver? (Consistent failure versus inconsistent failure)
Replies (1)
-
@phils@chaos.social 2026-05-02 08:20
@namedbird@noc.social @nygren@hachyderm.io I would not call it worse but much harder to debug. In a non-trivial case with CDNs and dynamic sets of DNS servers, you get a random looking set of DNS servers of which a random fraction has AAAA records. Depending on which you pick and how many failures you tolerate, this gives you a good chance of resolve failures. Checking 10sec later with another tool behaves totally different. Checking from another vantage point behaves totally different.