Re: [dnsext] Fwd: RFC 2308 & RFC 4035

[Mark Andrews <[e-mail]>]

In message <a06240800c98ed0f7a8c5@[10.31.200.114]>, Edward Lewis writes:
> At 11:32 +1100 2/26/11, Mark Andrews wrote:
> >Nothing worse than having asked for the AAAA record at 10:00 am.
> >The zone administrator has to assume that AAAA queries will have
> >been made prior to adding the AAAA records.  This will almost
> >certainly be the case if there were A queries.
> 
> I think my question wasn't understood.
> 
> Do cache implementers use the fact that the proof of no A record 
> coincidently shows there is also no AAAA record when processing a 
> subsequent AAAA query.  (Meaning, there's no other entry for the AAAA 
> alredy in the cache.)
> 
> If a cache does do this kind of proof by inference, what does the 
> implementer think about the situation in which there is a "race 
> condition" - that the proof of non-existence has been obviated by 
> actions outside of it's sphere?  (Like the subsequent adding of the 
> AAAA set in the authority.

We do this for NXDOMAIN today.  If we ask for A foo.example and get
back a NXDOMAIN then the cache gets asked for AAAA foo.example we
use that NXDOMAIN.  Using the NSEC/NSEC3 that matches the <qname,qclass>
really no different.

The much more interesting question is infering NXDOMAIN for a
<qname,qclass> we havn't asked for.  Named already does this when
looking for DLV records when validating but is expected by all
parties involved.

Mark
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]