On Wed, Apr 13, 2016 at 11:32:40AM +0000, bruno.decra...@orange.com wrote: > Could the draft clarifies the BGP Route Reflector behavior when reflecting a > route received from a buggy 3107 implementation with S=0? > a) One may argue that the RR should not modify the NLRI and hence reflect the > route with S=0 > b) One may argue that the RR should reflect the route after setting S=1 in > order to be a speaker compliant with 3107bis & 3107. > c) One may argue that this is an error and treat the IP Prefix as withdraw > > IMO I think I would prefer b) in order to not propagate the bug.
Normally I would argue for c. What I'm really curious about is implementations that may not be setting end-of-stack currently for their 3107. Time to read what our code does upon receipt. -- Jeff _______________________________________________ BESS mailing list BESS@ietf.org https://www.ietf.org/mailman/listinfo/bess