Re: [v6ops] A common problem with SLAAC in "renumbering" scenarios

[Richard Patterson <[e-mail]>]

On Tue, 26 Feb 2019 at 10:59, Fernando Gont <[e-mail]> wrote:

> >  If not, would an update to 6724 that introduces a Preferred Lifetime
> > comparison as a tie-breaker help?  i.e. When deciding between multiple
> > non-deprecated source addresses, always pick the address with the
> > highest Preferred Lifetime?
>
> Not really. If you were to follow that road, then fore each
> prefix/router you should keep a new "Last Advertised" timer and use that
> as a tie breaker.  Using the Preferred Lifetime would not be safe since,
> while there are default values, you don't really know what values the
> CPE will use for them.

I don't think that or anything else is required to change on the CPE
router.  The onus is on the end host to track the Preferred Lifetime
counter (it already does), and to adjust its source address selection.
  It doesn't really matter what the default values are, as long as the
CPE router continues to send out RAs with that PIO to keep the timer
refreshed.

Although what I do think might be an issue is in a multi-homed
environment, an end host might flip flop outgoing connections between
prefixes depending on which PIO was refreshed last.