IMHO, it is possible that a multihoming solution will end up requiring a message that informs the host about the correct address to use.
It would be good if we didn't have to define an additional message for this and we could just use this one. I think that Christian's proposal doesn't change the packet format, since the router just has to be smart enough to pick the right source address. So all that it is needed is to state that the router has to pick one of its own addresses that its ingress filters would allow to flow towards the selected destination. I fail to see why do you think this is not feasible Regards, marcelo > -----Mensaje original----- > De: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > Enviado el: viernes, 30 de enero de 2004 21:19 > Para: [EMAIL PROTECTED]; [EMAIL PROTECTED] > CC: [EMAIL PROTECTED]; [EMAIL PROTECTED] > Asunto: RE: ICMPv6: New destination unreachable codes > > > If I am not wrong, if we wanted the router to send the correct > prefix to the host in the ICMP message, we will have to change > the format of the ICMPv6 Destination Unreachable message. > > Currently, we just have 4 bytes unused in the message format and > this prefix can't fit in there. > > So the question is: is it worth changing the format of the message > for this optimization ?? > > Regards > Mukesh > > > -----Original Message----- > > From: ext Christian Huitema [mailto:[EMAIL PROTECTED] > > Sent: Friday, January 30, 2004 12:09 PM > > To: Pekka Savola > > Cc: [EMAIL PROTECTED]; Gupta Mukesh (Nokia-NET/MtView); > > [EMAIL PROTECTED] > > Subject: RE: ICMPv6: New destination unreachable codes > > > > > > > FWIW, I have no objection to specifying a new code, but I > > don't think > > > adding a "working address" option is feasible or useful. > > > > I understand that adding a new option is much harder than adding a new > > code point. I am fine with the new code point. > > > > > On Fri, 30 Jan 2004, Christian Huitema wrote: > > > > In a site exit scenario, ingress filtering is performed either at > > the > > > > ingress interface of a router, or at one of the exit interfaces on > > the > > > > router. I suggest that the source address of the router's ICMP > > message > > > > should be one of the global scope addresses associated to that > > specific > > > > interface. This gives a strong hint to the host: among the source > > > > addresses that can be tried, pick the one that is the > > best match for > > the > > > > router's interface. > > > > > > I believe that all router implementations pick the source address of > > > the generated ICMP error messages based on the outgoing interface of > > > the message: this would be toward the site's internal > > infrastructure, > > > and would very likely include addresses from all the prefixes. > > > > > > So this would probably not help in this specific case, > > unless you want > > > to make a specific exception (which might have some obvious > > problems). > > > > If the site is multi-addressed, the exit router link may or may not be > > multi-addressed. If it is not, then the global scope address probably > > matches the egress that the router is managing, and that is > > fine. If the > > router link happens to be multi-addressed, then it would be > > nice to pick > > the address that is most likely to match filtering. > > > > -- Christian Huitema > > -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
