Administrivia Scribe: Evi Nemeth (evi@cs.colorado.edu) Agenda Logical RP Draft Status -- Meyer Static Allocation Draft Status -- Meyer Allocation of Multicast Addresses in IPv6 -- Haberman Should MSDP not Advertise Some Groups -- Meyer MAESTRO BOF Summary -- Oran Inter-Domain Multicast Routing Issues -- Estrin Multicast Session Control Center -- Fasano Mzap Status -- Thaler Experiences with MSDP and MRM -- Hall Status of Internet Multicast -- Meyer/Fenenr Detailed Minutes Logical RP Draft Status (Meyer) Randy Bush: Concerns about how to debug problems. Tom Pusatari: Hard to debug in PIM v2, since BSR knows but nobody else knows. We need to know which box advertised an SA to you. Bill Fenner: Was the discussion about how to do the inter-domain version included as well as the intra-domain version. Need explicit discussion about inter- and intra-. feels there is an interaction. Dorian Kim: The inter-domain stuff was put in draft 0 because it might be a neat thing. having tried it, now know you don't want to do this. Dave Thaler: If we are going for bcp leaving it out is fine, but if we are going for informational, we need text that says why you should not do it. need the warning there. Randy Bush: Why invent new terms re. anycast. Meyer: were using the term Anycast, but Deering and others objected to its use. Bill Fenner: RFC 1546 seems to define anycasting in the way we are using it. A "hum vote": 2/3 for "anycast rp", 1/3 "logical rp". Carried. The name will be changed back to Anycast RP. Static Allocation Draft Status (Meyer) Few comments made. Should go to WG last call (experimental) soon. Meyer: Learned after the meeting that the IANA has allocated 233/8 for this use. Fenner: Add section to the document on transition and on getting the addresses back. Handley: If we use BGMP, these addresses are not aggregatable so how do they work in a GRIB in BGMP. Thaler: Doesn't think its a major problem Meyer: wants to update doc and get it going because lots of people are using static addresses and we need a doc to support them. Allocation of Multicast Addresses in IPv6 (Haberman) Use the ipv6 network prefix for a domain in multicast address Thaler: likes it for interface to routing, AS number not included in bgmp so its nice and simple. Fenner: one of the reasons to limit to lower 32 bits is to not collide with the ethernet link level address in the lower half. haberman's proposal has 48 bits available and wants to use them. but that messes things up. bill says you have already 64 bits of uniqueness. someone elsewhere (ipv6 wg's) made a decision that conflicts in the ethernet address were bad. Should MSDP not Advertise Some Groups (Meyer) SA caches get full of junk, maybe need to have BCP to describe filters you need to put on MSDP. RP's announcing SA's for internal sources. Hardware whose dhcp litters the SA caches, Thaler: what do you break if you filter, e.g., don't filter mtrace.mcast.net. You would have to look at each one. SLPv1 is the bad guy for SVRLOC. may want to coordinate with SLP working group before recommending filtering. Fenner: for the service discovery items, what TTL are these sent with. Folks assume TTL scoping. Might want something like "don't forward an sa unless you have gotten a packet with ttl > 16". Kim: Should write a doc that says dont forward this type of thing. Thaler: What is the relationship between these entries and those where only 1 packet was received. might fix it in MSDP to eliminate the 1 only packets. Fenner: Meyer's data shows that msdp infrastructure is not passing everything everywhere. His data shows 5k sa's and here are 6k bogus ones. Meyer: For example, a campus power outage adds thousands of these. Pusatari: Tell IANA not to allocate these, hunt down these people and tell them not to do it, people reuse addresses for different purpose because they are so hard to get.