[dns-operations] Missing .us and GTLD records??
vixie at isc.org
Fri May 21 05:17:38 UTC 2010
> Date: Thu, 20 May 2010 23:38:45 -0400 (EDT)
> From: Luis Uribarri <uribarri at alum.mit.edu>
> Yes, if there is a transport problem, then a dual stack resolver will have
> issues talking to a dual stack authoritative server, but it will not be
> compleately unable to talk to such authoritative server because of the
> availability of IPv4 on both the resolver and such authoritative server.
so you've got a resolver that behaves differently when some nameservers are
multihomed and some are not? i'd like to learn more about this, plz explain.
the only thing a resolver is SUPPOSED to do differently when encountering a
multihomed server is propagate SERVFAIL to the other addresses of that name
server, effectively poisoning all addresses for a "nameserver name" when it
finds a zone unresolvable there. you seem to think that your resolver is not
following RFC 1034/1035. what resolver is this? a correctly implemented
resolver will not give poorer performance for a singly home nameserver than
for a multihomed one, where all addresses for that nameserver give timeouts.
(timeout isn't propagated, so, the addresses aren't even considered related
except in the case where one of them returns SERVFAIL.)
More information about the dns-operations