OK -- let's try to summarize this.
1) let's avoid lengthy rat-holing discussions here,
2) a longer applicability statement might be good, but that's
independent of this document, and
3) it might be useful to add a short warning of QTYPE=* in this
document.
I propose adding the following as Section 1.4:
<section title="Query Type 'ANY' and A/AAAA Records">
<t>QTYPE=* is typically only used for debugging or management
purposes; it is worth keeping in mind that QTYPE=* ("ANY" queries)
literally return any available RRsets, not *all* available RRsets.
Therefore, to get both A and AAAA records reliably, two separate
queries must be made.</t>
</section>
Objections, rewording, ...?
On Wed, 5 May 2004, Edward Lewis wrote:
> At 7:57 +0300 5/5/04, Pekka Savola wrote:
> >Isn't the use of QTYPE=* causing more problems than its worth, being
> >unreliable and all that? Or is it's usefulness precisely restricted
> >to identifying what _is_ in the caches (and what is not)?
>
> This is an ancillary comment, based on the experience of being
> involved in DNSSEC and the longer-than-it-should-have-taken road it
> has travelled.
>
> I'd address this issue like:
>
> A) Don't bother discussing the usefulness of QTYPE=* here. It isn't
> a v6. (Debating the value of a feature good "only for
> troubleshooting" is a rat hole. There's no objective answer. Ever.)
>
> B) Write the guidelines for IPv6 adoption mindful of all the quirks of DNS.
>
> C) Don't smooth over the hard parts - doing so only creates the
> dreaded "corner cases." Overly strict "rules" eventually cause
> conflict. Instead, be loose enough to absorb the shocks of running
> over a rough road.
>
> DNSSEC at first tried to solve for only the normal cases of DNS, as
> well as some cases that were interesting to the security community
> (like alternative chains of trust). A lot of the delay of the
> security extensions to DNS came from not having enough of an
> understanding of the base protocol.
>
> To address the questions 1,2,3...
>
> 1) Personally I don't see QTYPE=* as an oddity, it's just not very
> clear to the causal observer. I'd not mention in the document
> anything positive, but a warning against assuming that QTYPE=* will
> retrieve both A and AAAA if they both exist.
>
> 2) I doubt there's much value is writing another document on QTYPE=*.
> It's defined in 1035 +/- 1, well enough I think. (It's actually
> QTYPE=ANY, which makes it easier to understand - it's not QTYPE=ALL!
> The latter is the assumption many make.)
>
> 3) No - but perhaps we (and I) ought to be clearer and never refer to
> QTYPE=*. There are times I rely on the ANY QTYPE for debugging.
>
> This is off-thread, but a similar situation holds for the truncation
> discussion. Ohta's better at it than I, the morale is not to dive
> into a treatise on the additional section for v6 without more
> research on the TC bit settings. There's a lot already written in
> the topic. I know - DNSSEC also messed with the rules there. ;)
>
>
--
Pekka Savola "You each name yourselves king, yet the
Netcore Oy kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
.
dnsop resources:_____________________________________________________
web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html
mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html
