Danny Mayer writes: > You did misunderstand. I said the Refid itself and not an extension > field so it would be interoperable. I'm of the opinion that a server > should only have one refid for all interfaces no matter which one is > being used to send the packet. I recall that there is an issue that > Harlan had to explain to me in excruciating detail as to why the > existing refid is the way it is today but I don't recall now the > details. We already have a problem where the hash of the IPv6 address > can conflict with a real IPv4 address.
If that would be your policy then the way to implement it is with the suggest-refid EF proposal I submitted. The legacy behavior is that "For S2+, I use the IPv4 (or partial hash of the IPv6) address of my system peer as the refid." If this situation is not clearly described in my proposal please show where that information could be improved. H _______________________________________________ TICTOC mailing list [email protected] https://www.ietf.org/mailman/listinfo/tictoc
