[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: WPD-6, WAA-8, and WAA-9 of draft-ietf-v6ops-ipv6-cpe-router-04

> part       text/plain                1708
> > WPD-6:
> >    Absent of other routing information the IPv6 CE router MUST
> >    use Router Discovery as specified in [RFC4861] to discover a
> >    default router(s) and install default route(s) in its routing
> >    table with the discovered router's address as the next-hop.
> >
> > RFC-4861 describes how routers process Router Advertisements (mainly to
> > check them for consistency). Only hosts maintain a list of default routers.
> 
> that's why the upstream interface is specified as behaving as a
> host for some functions. see W-1.

For acquiring a global address. It doesn't say anything about after that. Does
it remain a host until a default route is installed? How does it handle 
expiring the router router. Or is the first router that replies installed there
forever?

> > It looks like that at least one ISP treats this draft BCP as gospel. So I
> > wonder whether the draft should mandate the weak host model. I can't think
> > of anything negative that would happen if a CPE assigns in WAA-8 the addres
> s
> > not to a virtual interface but to its WAN interface and then implements the
> > strong model.
> 
> that would violate rfc3633. which require that a delegated prefix
> isn't used on the link between the RR and the DR. routers are
> generally following the weak host model, any particular reason why
> you require the strong host model?

I never expected a DHCP RFC to effectively mandate the weak multi host model.
Weird.

Because the CPE is essentially always-on it is very tempting to add other host
functionality, like a file server, a web server for the customers's pictures
and videos, maybe a SIP gateway, etc. As such it may acquire other prefixes
as well, maybe ULA, maybe as part of multi homing, prefixes from other
providers.

With the strong host model, you know from with interface addresses are going to
be selected for sending error ICMPs. With the weak multi homing model it is
anyones guess.

How do prevent to router from using the wrong address for sending an error
ICMP?