On Thu, Dec 5, 2024 at 6:04 AM Luc Andre Burdet (lburdet) <lbur...@cisco.com> wrote:
> The normative SHOULD was used because it is possible to operate in > Port-Active mode without the P/B in Ether A-D per ES (if port-active is > setting the interface Down then the peer will not send any routes – there > is no ‘B’). > Would someone ever choose to do that? If so, why? If this is just for backward compatibility, there are other alternatives like saying new implementations MUST do (new thing), and SHOULD accept (old thing) for backward compatibility. It is just undesirable for convergence: the P/B give a hint to the remote > that the entire port is Primary vs Backup, and that it need not wait for > each individual Ether A-D per EVI update to P/B. > That sounds a lot like a MAY to me. > All implementations I am aware of are doing it, but I should still circle > back to co-authors/implementors before flipping it to a MUST. > OK, sounds good. -MSK
_______________________________________________ BESS mailing list -- bess@ietf.org To unsubscribe send an email to bess-le...@ietf.org
