On Fri, Dec 05, 2025 at 02:52:28PM +0100, Matteo Perin via Bird-users wrote: > Dear BIRD dev-team and community, > > This is a patch proposal spawned from the discussion in a previous mailing > list thread [0], regarding how to achieve automatic BGP peering establishment > using the IPv6 Neighbor Discovery mechanism. > > Over the last few weeks, I worked on implementing the feature with the > outline Maria Matejka kindly provided. > As expected, it proved to be quite challenging, especially when it came to > integrate the peer discovery approach with the current dynamic BGP > implementation.
Hi Thanks for the patches. From cursory reading these patches look good. I have several comments: - I think that peer net_addr schould contain interface (in the form of interface id). That would eliminate need for 'from' option in BGP and also simplify config in case of link-local neighbors on multiple interfaces. Alternatively, one could consider sending interface as a route attribute (like in next hop attribute for regular IP routes) but i think in this case having it as a part of net_addr would make more sense. - peer channel in BGP should be regular struct channel (like MPLS channel) and not struct bgp_channel. This is for channels that are also AFI/SAFI on the BGP session. 'from' option is not necessary as above, 'persist' option could be BGP option. - It seems to me that option 'neighbor discovery' should be called more like 'router discovery' to avoid confusion with detecting regular IP neighbors by listening to Neighbor Discovery messages. - I would consider to set pxlen for peer_net_addr as 0 in the secne that the field is unused. -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: [email protected]) "To err is human -- to blame it on a computer is even more so."
