Editor's note: These minutes have not been edited. The OSPF Working Group met on Wednesday, June 26th from 1930-2200 at the Montreal IETF. Minutes of the meeting follow: (1) John Moy explained the change in the most recent OSPF Version 2 I- D. The draft did not make the I-D submission deadline, but can be found at ftp://gated.cornell.edu/pub/ospf/draft-ietf-ospf-version2- 07.txt. The change fixes a possible routing loop when the same AS- external- LSA is advertised from multiple areas, with at least one using a forwarding address. Dennis Ferguson pointed out that one unfortunate side effect of the change is that the precise ASBR that will be used can no longer be predicted by the sender. (2) John then summarized the OSPF for IPv6 draft, and explained the changes from the last version. Possible future changes to the draft included: a) Link-LSAs generated only on links having associated network-LSAs, b) the priority field may be going away in IPv6, changing the OSPF encapsulation, c) IPv4-compatible addresses may be banned from propagation in OSPF (see below) and d) there may be some restrictions on where link-local IPv6 addresses can appear in LSAs. (3) Ran Atkinson gave a presentation on propagation of IPv4- compatible IPv6 addresses within IPv6 routing protocols. Ran also gave this presentation in the IPng WG (first session). Ran was asking that to avoid unnecessarily increasing the size of a router's IPv6 routing table, propagation of IPv4-compatible addresses should be banned. There seemed to be general confusion as to the use of IPv4- compatible addresses. Bob Hinden said that rather than put this restriction inside the individual routing protocol specs, a separate document was needed to apply the policy to all routing protocols uniformly. (4) Sandra Murphy gave an update on "OSPF with Digital Signatures". MaxAge LSAs are covered by the signature, while otherwise the LS age field is omitted from the signature. This prevents a router from flushing other router's LSAs; incrementing an LSAs age to close to MaxAge does no harm due to OSPF's MaxAgeDiff rules. In OSPF for IPv6, Digital Signature support may be able to make use of IPv6's flow ID, the new OSPF "flood even if not understood" option, and the Instance ID for running multiple instances of OSPF on a single link. Digital Signatures do not remove the need for MD5 packet-based authentication; both are needed for real security. The downside of Digital Signatures is increase in database size, and that the signing and verification takes a lot of router CPU cycles. "OSPF with Digital Signatures" has been implemented within a simulator, and is being ported to GATED and possibly the BBN Gigabit router. (5) Ray Nair and Bala Rajagopalan gave a presentation on the requirements for QoS routing. They stressed that QoS routing is more [Page 1] than adding metrics to OSPF -- many efficiency and scaling issues need to be dealt with. For example, tree-based flooding of frequently changing information or metric quantization to reduce update frequency. (6) Eric Crawley gave a talk on QOSPF, an extension of OSPF and MOSPF for QoS routing. Currently QOSPF deals only with source/destination flows (no protocol number). QOSPF uses the routing to actually signal the path for a connection; one reason this was done within the routing was to avoid changing RSVP. Talk ended with a discussion of motivation behind QoS routing, with the following possibilities: "accommodate more flows", "give varying level of service to packets", "improve chance of accepting flows" and "finding paths that wouldn't necessarily be found". At the end of the meeting it was mentioned that the implementation of OSPF MD5 authentication in GATED does not match the current OSPF specification.