Re: Adoption Call for "Improving the Robustness of Stateless Address Autoconfiguration (SLAAC) to Flash Renumbering Events"

[Michael Richardson <[e-mail]>]

I think that we have a rough consensus that there are events that seem to
cause long-connecting devices (such as desktops, TVs, appliances) to
sometimes fail to notice when there is a renumber.
("Short-connecting" devices like phones tend to join/leave often enough that
they usually get recent info. and they get tested more for this)

It seems that for the non-flash case, that there is a lack of complete
testing/implementation for the various cases, most of which are probably
already dealt with given proper implementation of the protocols.
The result is that many of us see renumbers which do not get deployed
smoothly, and we assume a flash renumber when in some cases, it was not.
("Your failure to plan ahead is not an emergency to me")

The result is that the urgency of this situation varies by experience, and by
quality of implementation.  It seems that if we had an RFC that dealt with
the **FLASH** renumbering situation, that by construction, devices that dealt
with that would also deal smoothly with the planned renumering situation.
(Didn't we have an entire WG at one point to renumbering? I think it didn't
end that successfully, but that was during years I spent less time on IETF)

As Erik points out, I think that there a number of active measures that can
mitigate the situation without changes to RAs, but I also think that we can
plan long term such that each end of the conversation is trying to make
things go smooth.

To clarify: I think we should adopt this document, provided we can more
constructive understand that we do not all see the same part of the Elephant.

--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-