[EMAIL PROTECTED] (Pekka Savola) writes:

> As a generic observation: the document seems to include some of much the
> same material as draft-ietf-dnsop-ipv6-issues-11.txt (specifically, its
> appendix B).
> 
> If folks think it would be a good idea, some or most of the
> stuff in that draft's appendix could be moved here -- as it has a lot of
> already IESG-reviewed material on additional section processing.
> 
> Notes below show some issues which also came up (in one form or another)
> during dnsop-ipv6-issues.  This leads me to think the text needs significant
> edits and/or merging/copying form dnsop-ipv6-issues appendix B.

the draft name is draft-ietf-dnsop-ipv6-dns-issues-11.txt for those who,
like me, try to use google to find its text.

i don't nec'ily consider the iesg an expert on this, so their approval
may or may not matter much.  the current -respsize- draft describes what
a large part of the installed base of dns query initiators actually does.

the best thing would be to shorten -respsize- by removing anything it
does not absolutely HAVE to say.

>     2.2. If the total response size would exceed 512 octets, and if the data
>     that would not fit belonged in the question, answer, or authority
>     section, then the TC bit will be set (indicating truncation) which may
>     cause the requestor to retry using TCP, depending on what information
>     was desired and what information was omitted.  If a retry using TCP is
>     needed, the total cost of the transaction is much higher.  (See [RFC1123
>     6.1.3.2] for details on the protocol requirement that UDP be attempted
>     before falling back to TCP.)
> 
> ==> "which _may_ cause the requestor..." is incompliant with RFC2181, which
> requires that if TC is set, the whole response must be discarded, and then
> you retry with TCP.  (Whether or not that's efficient or robust is a
> different argument, I guess, but that's what the specs mandate.)

forgive my ignorance.  i was describing behaviour i'm aware of.  if rfc2181
says that, it's boneheaded, but we should follow it here (or at least remove
any text in -respsize- that conflicts with it) rather than argue about it.

>     2.3. RRsets are never sent partially unless truncation occurs, in which
>     case the final apparent RRset in the final nonempty section must be
>     considered "possibly damaged".  [...]
> 
> ==> this may require a bit of rewording, as above, as the spec says to
> discard the response if TC was set no matter what.
> 
>    2.6. Average and maximum question section sizes can be predicted by the
>     zone owner, since they will know what names actually exist, and can
>     measure which ones are queried for most often.  For cost and performance
>     reasons, the majority of requests should be satisfied without truncation
>     or TCP retry.
> 
> ==> as above, "truncation or TCP retry" should maybe be "truncation and
> TCP retry"
> 
>     2.9. The best case is no truncation.  This is because many requestors
>     will retry using TCP by reflex, or will automatically re-query for
>     RRsets that are "possibly truncated", without considering whether the
>     omitted data was actually necessary.
> 
> ==> as above, not having such "reflexes" would be against RFC2181

and yet, since the installed base actually works the way this -respsize-
draft describes, the issues that will affect choices of NS RRsets are
exactly as written, and rfc2181 is possibly-irrelevant.  perhaps adding
a small editorial indicating this schism is the best fix overall?
-- 
Paul Vixie
.
dnsop resources:_____________________________________________________
web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html
mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html

Reply via email to