[prev in list] [next in list] [prev in thread] [next in thread]
List: djbdns
Subject: Re: Weird behaviour with ANY query to dnscache
From: Henning Brauer <lists-djbdns () bsws ! de>
Date: 2001-12-30 13:08:49
[Download RAW message or body]
On Sun, Dec 30, 2001 at 01:16:25AM -0500, Scott Gifford wrote:
> Henning Brauer <lists-djbdns@bsws.de> writes:
> > On Sat, Dec 29, 2001 at 01:09:03PM +0000, Jonathan de Boyne Pollard wrote:
> > > The root cause of the problem is that the DNS client library linked
> > > into "qmail-remote" is passed a 512 byte buffer in which it returns its
> > > results, and this buffer is simply not big enough for some answers.
> > WRONG.
> > The root cause of the problem is the existance of > 512 bytes answers for
> > such queries. There is no sane reason to do so.
> Hrm...I'm not sure I agree with this, Henning. The DNS specification
> clearly allows for it, even goes out of its way to describe how to
> handle it. It doesn't break any rules I'm aware of. It does require
> a TCP connection (which is obviously inconvenient),
In short: it requires major overhead. It is not only a tcp query with the
associated handshake, it is also a re-query - the udp query was long done.
> and breaks qmail
> (which is even more inconvenient), but that doesn't make it
> automatically insane.
right. Point is: there is simply no reason to need more than 512 bytes for
the MX list.
> Can you provide pointers to any standards that say that large answers
> should be avoided in all circumstances, or that is acceptable for DNS
> clients to malfunction if answers are larger than 512 bytes?
No.
That's why I said "there's no reason to do so", and not "this breaks RFC
xxx" ;-)
Greetz
Henning
--
* Henning Brauer, hostmaster@bsws.de, http://www.bsws.de *
* BS Web Services, Roedingsmarkt 14, 20459 Hamburg, Germany *
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic