On Thu, 3 Dec 2009, Ole Troan wrote:
with regards to source address selection. it might be prudent to state that the IPv6 CE router supports the weak host model and can use a source address from a different interface than the interface the packet is sent out of.
I agree, I think it would be nice to mention that in the document, I think it deserves a clarification.
Then the wording here: "WPD-3: If the delegated prefix is an aggregate route of multiple, more-specific routes the IPv6 CE router MUST discard packets that match the aggregate route, but not any of the more- specific routes. In other words, the "next-hop" for the aggregate route should be the null destination. This is necessary to prevent forwarding loops when some addresses covered by the aggregate are not reachable. [RFC4632]" Would it make sense to specifically say that any PD space received must be null routed unless it's used on an interface or routed somewhere else (sub-PDed to another router in the home)? Just in case multiple PDs are received and a LAN interface goes down?that's what we are trying to say. i.e that you have a null route for the aggregate, aka the delegated prefix. parts of this text is from rfc4632. any improved proposed text would be appreciated.
Darn, I've just spent 10 minutes trying to come up with better wording but I can't really come up with something better that is still simple and short. Let's leave it as it is, at least I can't come up with something better.
"L-5: The IPv6 CE router must assign a separate /64 from its delegated prefix (and ULA prefix if configured to provide ULA addressing) for each of its LAN interfaces. The IPV6 CE router MUST make the interface an advertising interface according to RFC4861. In router advertisements messages, the Prefix Information Option's A/L-bits MUST be set to 1 by default; the A/L bits setting SHOULD be user configurable."Does "LAN interface" refer to physical or vlan interface? What if the home CPE has 8 LAN ports and the user would like port 1-3 to be in the same vlan and thus share the same /64. Does this wording prohibit this behaviour? There is a "must" (shouldn't that be capital letters?) in there... I think the behaviour of having multiple ports in the same /64 and do L2 switching between them should be allowed.we have received multiple comments on this. suggestions are welcome! since this is a router I thought it obvious that whenever we talked about an interface it would be clear that it was a L3 routed interface connected to an IPv6 link. this document does not concern itself with bridging or switches on L2. e.g a typical IPv6 CE router would have 4 downstream Ethernet ports, but these are typically switched and represented as a single IPv6 LAN interface.
What about adding text to L-5:"... its LAN interfaces (single logical interface, might be one or multiple physical ports with L2 switching between them). ..."
But if we go into this area, should we make recommendations regarding MLD snooping in switched L2-domains as well?
-- Mikael Abrahamsson email: swmike@swm.pp.se