Tim Chown wrote:
On Tue, Oct 28, 2003 at 09:59:20AM -0800, Christian Huitema wrote:
It worked fine when we ran it at IETF51 in London.
You do mean DHCPv6 for DNS configuration, right? Many platforms do not support address assignment via DHCPv6...
I think the idea is for DNS resolver discovery... but even less probably support that?
I get your point about current implementation support but think it's worth discussing some issues.
I think that one of the big things about Router Discovery and Stateless address autoconfiguration is that people assume that further configuration mechanisms aren't needed, and that RA options are the correct place for additional config information.
While the addressing and routing are handled in RA reception and subsequent SAA operations, there's an entire bit ('O' Flag) in the RA header which says consult your DHCP server for further configuration.
This actually implies that implementations of stateful addressing and other configuration may be separate in some way.
Principally, if M=0, the other configuration doesn't need to be handled statefully on a server (or may be cached on the relay/router).
This means the implementation of DHC on routers doesn't need to be terribly sophisticated, and avoids lumping generic configuration into a service which is addressing and routing oriented (Router Advertisement).
Greg Daley
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------