Re: [v6ops] draft-ietf-v6ops-dhcpv6-slaac-problem

[Brian E Carpenter <[e-mail]>]

On 12/01/2017 22:51, Fernando Gont wrote:
> Hi, Ole,
> 
> On 01/12/2017 05:45 AM, [e-mail] wrote:
>>> This draft describes real issues but the information is a bit out
>>> of date. Can we ask the authors to confirm that things are still
>>> the same for current o/s releases? Then publish quickly so that we
>>> can start the *real* discussion: how to fix?
>>
>> This draft shouldn't be gating any real discussion. If there are
>> solutions please offer them now.
> 
> Noting the problem, and providing datapoints is useful, IMO.
> 
> 
> 
>> I have read through the draft again. While the draft has some points,
>> these are issues that has been discussed extensively. I struggle to
>> see how this will lead to and "fixes".
>>
>> I have a suggestion. Given that we have beaten on this particular
>> horse a lot over the last 15 years. Could we redo this draft into a
>> "Why?" draft in the style of RFC7421? Explain the issues, explain the
>> design choices and why the mechanisms function as they do, but don't
>> phrase it as a problem that needs to be fixed.
> 
> The fact that you get different outcomes for different systems, setups,
> etc., does not seem nice to me.

Yes, that *is* the problem. And, while I agree with Ole's points about how
to improve the document by making it more explanatory, I think documenting
the problem properly is necessary before proposing fixes.

> As to the why...I wonder if a "well, there's the slaac folks, and the
> dhcpv6 folks..." document would be "publishable"..

That's a *different* document, and I think the issue there is what Randy
Bush mentioned recently on another list: we don't offer feature-equivalence
between DHCP and DHCPv6, but sites that rely on DHCP operationally want
feature-equivalence, and they are the customers, and customers are always
right.

    Brian