Date: Mon, 14 Oct 2002 14:43:41 +0200 (CEST)
From: Erik Nordmark <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
| How is this intended to work with LLMNR (draft-ietf-dnsext-mdns-12.txt)?
| One way of using LLMNR is to make *it* be the last resort, but since
| we can't have two last resorts this is quite problematic.
LLMNR isn't really intended to be "last resort" in the same sense.
The "well known address" system can only be defended in way at all,
if you only ever use it (attempt it) if everything else has totally
failed. That is, rather than giving up and saying "network is
broken, I can't look up that name", a resolver can try this one last
(valiant but probably useless) attempt to get a reply from somewhere.
| Also, looking at the dns-discovery draft in isolation, it is far
| from clear to be how one can operationally control this behavior.
That's not surprising, one cannot, or it would be useless. Or perhaps,
more correctly, it's useless anyway, but if there was an easy way to
disable it (aside from providing some other mechanism which actually
works, which will effectively disable it), it couldn't even hold out the
false hope.
The real problem about the draft is that it is stealing my address space
to allow others to neglect rational configuration.
I agree with all Rob Austein said about this as well (certainly including
the issue of terminology), with the possible exception of ...
Finally, I remain unconvinced that this is a problem that the IPv6 WG
ought to be trying to solve at all.
Unless a solution emerges from somewhere else, I think this WG does need
to work on the problem, as one of the promises of IPv6 is stateless
autoconfiguration. And without being able to use DNS names, no matter
how configured the box believes it is, it is useless. Some workable
solution to this is needed. Personally I don't think this draft is a
satisfactory answer. It is more of a "well, we did this, and so the
problem can be handled" type answer, just so there is something. But
if that's what we really want, just for the IPv6 WG to provide something,
and then to assume that a real workable answer comes from elsewhere,
then, OK I guess.
I know I'll never be running recursive servers on the addresses it
specifies (most likely will never be using those addresses at all, but
who knows, it would be a good place to stick a server that simply
returns a NXDOMAIN reply to every query it receives) but given that
people insist on attempting to define what I'm supposed to do with the
address space allocated to me to manage, this is no worse than any of
the rest of that.
kre
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
