Pekka Savola wrote:
I'm not sure if I understand your comment. Are you saying the ND proxy
spec is too complicated?
Well, I myself suggested removing the spanning tree loop prevention from
the draft completely (now it has a bit in the RAs) because it wasn't
needed in the applicability we had in mind. But folks that didn't like
ND proxy argued that infinite loops are not nice, even in illegal
configurations, so we're stuck with some additional specification.
How would you like to see it simplified? Do you have an alternative in
mind?
(To me, ND proxies seems "as simple as it can get" excluding loop
prevention which I argued for removal but folks thought the failure
modes were too dangerous..)
The suggestions on the IPv6 list in the past was to keep the bit in the
RA so that complex topologies (which might have loops) can be detected,
but not include any attempt for the proxies to work when there is a
loop. Thus a proxy would not use downstream interface where there is
another proxy advertising RAs with the bit.
That would be sufficient to handle the scenarios in the document.
Erik
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
