Erik, > -----Original Message----- > From: Erik Nordmark [mailto:erik.nordm...@sun.com] > Sent: Tuesday, December 08, 2009 9:18 AM > To: Templin, Fred L > Cc: Narasimhan Venkataramaiah; IPv6 Maintenance WG > Subject: Re: Question on RFC 4191 inconsistency > > Templin, Fred L wrote: > > >>> The presence of default routes means that the host should > >>> accept packets from all four routers equally, i.e., and not > >>> only accept the ones from R1 and R2 while dropping those > >>> from R3 and R4, right? > >> What do you mean by "accept"? The content of the default router list (or > >> a routing table in general) doesn't have any impact on what received > >> packets are accepted by a host. > > > > I mean that on some link types hosts need to be careful > > about what routers they accept packets from, and should > > not accept packets from a node pretending to be a router. > > That might make sense in some cases. (But more typically it is routers > and other network devices like bridges that might want to verify source > addresses etc.)
I can live with that if you can, i.e., only have routers and/or bridges verify source addresses, and let hosts accept packets that may come from any routers. It would make life a lot easier for the hosts. > But I don't see what that has to do with RFC 4191. > > > Some routing protocols declare a link down when a certain > > number of hello messages are lost. I don't see why NUD > > could not be used in the same way. > > Perhaps because NUD is *NOT* a Hello protocol. > With enough thrust pigs can be made to fly, but that doesn't mean it is > a good idea. This is not an all-pigs-flying concept. As long as the route lifetime has not expired, if NUD can be used as a hint that the route has become non-operational then why can't it be used as a hint that the route has come back to life. > The closest thing we have to a hello protocol in ND are the periodic > RAs. And when the default router lifetime from the RAs time out, then > the default route, as well as any more specific routes from that router, > will time out. Correct; a route that has timed out is no longer usable. Nothing about this changes that fact. But, by my read of the spec the router lifetime applies only to the node's ability to act as a default router - not its ability to service more-specific routes. Fred fred.l.temp...@boeing.com > Erik -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
