On Fri, 2014-02-28 at 14:05 -0500, Michael Richardson wrote:
> Paul Wouters <[email protected]> wrote:
>     > 1 Applications can either do dnssec validation themselves, or trust the
>     > AD bit.
> 
>     > 2 It is undesirable that each application has its own DNSSEC validation
>     > code, trust anchors and DNS cache.
> 
>     > 3 It is undesirable that applications blindly trust the AD bit when
>     > resolv.conf points to another host as the AD bit could have been 
> modified
>     > on the network.
> 
>     > 4 In the ideal world tomorrow, each host has its own automatically
>     > configured, perfectly working validing DNS server and resolv.conf can
>     > be ignored or is always hardcoded with nameserver 127.0.0.1
> 
> My problem isn't that the AD is insecure, but that it isn't very useful.
> Going back 10 years to the various DNSSEC workshops, one of the things that I
> wanted was more information about why there was a validation failure.
> 
> For instance, if I have previously contacted example.com, and I have
> it's A/AAAA or more interestingly, the DANE borne public key for the service
> I want to reach cached, or leap-of-faith'ed, I don't care as much if the
> DNSSEC fails to validate because a signature expired.
> 
> If it fails to validate because the data is correct, I expect the bad data to
> be discarded, and for it to try again.  At some point (<<5s) the application
> needs to get some kind of report that name is not presently available.
> (Happy eyeballs, or some other mechanism might want to try something else)
> 
> This is doubly true if I have contact with the user who
> can I can:
>     a) advise of the specific reason for the failure
>        (which up to now, would be followed by facepalm and one of geeks
>        goes to fix the problem....)
> 
>     b) find out what they want to do now.
> 
> SERVFAIL / "Host now found" is simply not acceptable information.
> 
> For this reason, I think that applications should not set or depend upon the
> AD bit, even if the resolver is ::1.  They either understand DNS(SEC), or
> they use an API call way more sophisticated than getaddrinfo() to do their
> connections.   Java had the right idea, but the implementation and error
> reporting was very poor.

Nothing in this proposal prevents you from doing that for applications
you care about. OTOH forcing applications to a completely new API by
refusing this proposal on your grounds will guarantee less applications
will use DNSSEC. And DNSEC support will rapidly fragment making
system-wide management a lot more difficult. I think that prospect is a
much worse evil.

Simo.

-- 
Simo Sorce * Red Hat, Inc * New York

_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane

Reply via email to