Re: [v6ops] Clarification/addition on the cpe-slaac doc.

[Fernando Gont <[e-mail]>]

Hi, Bernie,

On 10/2/21 12:37, Bernie Volz (volz) wrote:
> It seems odd to me to be cherry picking things for RFC-8415 to reiterate 
> as requirements?
> 
> Doesn’t that open up the possibility that someone skips other things 
> because “well it wasn’t in the CPE-SLAAC” requirements?

This was motivated by the default behavior of some popular CE Routers, 
which end up triggering the use of dynamic prefixes, because they simply 
randomize the IAID in such cases. So this is relevant here.

(One may wonder whether they are doing so by extrapolating RFC7844, or 
simply because they don't store the IAID on stable storage and also 
don't use an algorithm that always computes the same value...)


I think one could add this note to the corresponding subsection that 
explains the requirement:

    [RFC8415] requires that IAID for an IA MUST be consistent across
    restarts of the DHCP client. However, some popular CE Routers are
    known to select a new random IAIDs e.g. everytime the underlying PPP
    session is established. This could be the result of extrapolating the
    behavior described in [RFC7844], or simply a consequence of not
    storing the IAID on stable storage and failing to employ an algorithm
    that consistently generates the same IAID upon reboots. Thus, this
    requirement prevents CE Routers from inadvertently triggering a
    flash-renumbering event on the local network.


... or the like.

Thoughts?

Thanks!

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