Re: [spring] IPv6 EH-insertion (Re: Spirit and Letter of the Law (was: Question about SRv6 Insert function))

[Fernando Gont <[e-mail]>]

On 8/9/19 14:55, Michael Richardson wrote:
> 
> Fernando Gont <[e-mail]> wrote:
>     >> 
>     >> Why do these kind of arguments emerge right now instead of 5 years
>     >> ago?  We left the “problem “ for 5 years? And suddenly we notice them?
>     >> How interesting.
> 
>     > One possibility: All this kind of arguments did emerge a long time ago
>     > -- NAmely, when working on rfc2460bis, which eventually became
>     > RFC8200. As a results of the problems associated with EH-insertion,
>     > there was IETF consensus to explicitly ban it in RFC8200. Maybe some
>     > 6man'ers were not subscribed to the spring wg list, were quite
>     > surprised to see documents relying on EH-insertion (one ins spring, and
>     > another, indirectly, in lsr), and hence decided to comment.
> 
> I would suggest that the result was slightly different.
> 
> We decided in RFC8200 that EH-insertion by random devices for randomly
> defined purposes was banned.
> 
> That if a device/operator needed to do that, that they ought to use IPIP
> insertion. We failed to actually follow through with RFC8504 to make
> it possible to actually use IPIP in a general case.
> 
> What I think that we wound up in RFC8200, and I think that it was
> intentionally very subtly under-stated by the AD at the time, was that any
> IETF Specification could in effect make any rule it wanted, and it
> would be subject to IETF Last Call and IESG review.

Yeah, that's IETF process, right? You can make any rule that you want
(assuming you do it within a wg with the proper scope), and you get wg
review and ietf-wide review. No new here. And I don't know why the rules
should be any different for eh-insertion.




> So what we banned was EH-insertion by vendors or operators without
> significant (standards-level) thought as to applicability.

Huh?

EH insertion is banned. If you want to violate that, you have to publish
a spec that updates RFC8200. Either to remove the corresponding text in
RFC8200, or at least to include an "UPdates: RFC8200" tag, such that
folks that read RFC8200 know what are the excemptions to the rule.


> We are now here,
> we are supposed to be discussing if this use case, along with it's
> applicability, warant a reasonable exception.

I assume by "we" you mean IETF. Because this doesn't look like a minor
modification to IPv6. Hence, that would be out of scope for 6man.

Thanks,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492