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

[Lorenzo Colitti <[e-mail]>]

On Mon, Jul 13, 2020 at 9:47 AM Michael Richardson <[e-mail]>
wrote:

>     > I think it's very instructive to compare that process with the
> advancement
>     > of the IPv6-only DHCP option. That option essentially solves the same
>     > problem, but started off with a lot more consensus and has already
> gone
>     > through both working group last call and IETF last call.
>
> I think that we could never have found the IPv6-only DHCP option solution
> had
> we not first gone through the hundreds of emails.
>

Actually I don't think so. I think the problem we faced is that once a
document is adopted it becomes a WG document and the sunk cost fallacy
prompts participants to continue to work on it. If instead we had decided
that there was not enough consensus to adopt the IPv6-only flag, then the
authors could have looked for other solutions. As it happened the solution
that ended up being more successful wasn't even in this WG, it was in dhc.

The problem with your approach is that basically you are creating an
> informal
> (adhoc) working group before the working group.
>

Why do you say that? We're conducting this discussion on the WG mailing
list, are we not?


> So, to summarize:
>  1) you agree that there is a problem!
>  2) you like 70% of the document
>  3) once adopted, you want to seek consensus of the WG to remove section
> 4.5.
>
> So, you are in favour of adoption.
>

I think you are confusing lack of objection with lack of support.
Personally, I don't think this is primarily a standards problem but an
operational one, and I don't think the problem is widespread/important
enough to work on it in 6man (whose charter is to maintain/evolve the IPv6
standards). But I could be wrong about that, and don't oppose the WG
working on the problem in 6man if that's what folks want to do. Similarly,
I don't oppose most of the document, though as I said, almost all of that
is tweaks to the defaults that could be in a v6ops BCP instead of requiring
a standards change. I do strongly oppose the idea that we need a big change
to the mechanics of SLAAC to solve this problem, and oppose the WG taking
on the document on those grounds.