Re: [dnsext] DS digest downgrade

[Mark Andrews <[e-mail]>]

In message <1300940324.2117.247.camel@localhost>, Matt McCutchen writes:
> On Thu, 2011-03-24 at 14:51 +1100, Mark Andrews wrote:
> > The time you get a failure is if there are differing DS algorithm
> > sets for the same DNSKEY algorithm and you are pre-publishing DS
> > and the pre-published DS has a SHA256 DS when the current DS for
> > that algorithm doesn't have a SHA256 DS but has a SHA1 DS.  This
> > will leave you with a SHA256 DS pointing to a non-existing DNSKEY.
> > 
> > If one is not pre-publishing DS records I don't see a failure case.
> 
> Right.  Since an entire set of DNSKEYs is published, there are more
> chances for things to work correctly, compared to DANE where the TLS
> server presents just one of the designated certificates.
> 
> > Treating the zone as insecure is not a failure.
> 
> I disagree.  People are beginning to use DNSSEC to store opt-in
> indications for application-level security mechanisms associated with
> DNS names, such that treating the zone as insecure and allowing the
> indication to be removed would be a security failure.  You might think
> this is a bad idea, but the benefits of these mechanisms are compelling
> and I think it would make more sense to beef up DNSSEC to support them
> than to set up a parallel system.
> 
> What case are you thinking of in which the zone would show up as
> insecure?

Alg A has SHA1,  ALG B has SHA1 + SHA256.  Once you filter the SHA1
DS out you are left with just ALG B.

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