Re: [DNSOP] ANAME in answer or additional section [issue #62]

[Tony Finch <[e-mail]>]

Matthijs Mekking <[e-mail]> wrote:

> The main argument for putting it in the additional section is that given
> the experience with DNAME, putting the ANAME in the answer section there
> is a risk of interop problems (because there is an unexpected record in
> the answer section).

I think ANAME will cause too many problems if it puts unexpected records
in the answer section. Speaking as an ANAME proponent, the reason ANAME
is such a hack is for compatibility with the installed base, and it will
be annoying if it isn't actually compatible and we have to wait another
10+ years to be able to use it without worries.

We (mostly Chris Thompson) deployed DNAME in the reverse DNS in 2010 (the
DNAME specification was published in 1999) and we observed at least two
annoying interoperability problems:

* glibc chattering noisily in syslog (fixed only 2 years ago)
https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=b9b026c9c00db1a1b5b4a3caa28162655a04a882

* mail delivery failures - MTAs typically have their own DNS message
handling code which is often super careful

I expect that there will be several more interestingly problematic
DNAME failures in the forward DNS.

Tony.
-- 
f.anthony.n.finch  <[e-mail]>  http://dotat.at/
Gibraltar Point to North Foreland: Variable, mainly southwesterly 3 to 5,
occasionally 6 in south. Smooth or slight, occasionally moderate in south.
Showers then fair. Good, occasionally moderate at first.