Re: A common problem with SLAAC in "renumbering" scenarios

[Mark Smith <[e-mail]>]

On Mon., 4 Feb. 2019, 04:14 Mikael Abrahamsson <[e-mail] wrote:

> On Mon, 4 Feb 2019, Mark Smith wrote:
>
> > A first hop CPE rebooting and being given a different PD prefix is
> > effectively changing a transient packet loss event into the movement of
>
> What about if it decides to choose a different /64 out of that /56? Do we
> cover this in any documents? The result to end systems is the same
> regardless if it's different /64 out of the /56 or if it's different /56.
>

Are there ones doing that? The effect would be the same. Hosts are
effectively having their network location changed because the only valid
available prefix(es) is/are new one(s).

I'm doing a well overdue (on my part) review of rfc4941bis. Privacy
addresses are in some regards effectively doing this too, because
applications' transport layer connections cannot persist longer than the
privacy addresses' valid lifetime.

The difference is that privacy addresses are expected to have a minimum
valid lifetime that is larger than the longest expected transport layer
connection duration that uses a privacy address.

Regards,
Mark.


>