[lacnog] Articulo: Understanding why IPv6 renumbering problems occur

[Fernando Gont]

On 18/9/19 03:07, Fernando Frediani wrote:
> On 16/09/2019 07:16, Fernando Gont wrote:
>>
>>> I seems to me it would be more productive if somehow SLAAC could, one it
>>> is aware there was a prefix change to send a broadcast message with some
>>> kind of flag which devices should understand and fully refresh their
>>> IPv6 addresses immediately.
>> Well, an RA with proper lifetimes could, to some extent, achieve that.
>> But I believe for robustness-sake the smarts should be on  the hosts, as
>> opposed to the network.
> 
> In the situation described I think lifetime would be irrelevant, because
> once the CPE finds out it must use another PD it must inform all devices
> in all LANs and they should react to it immediatlly by changing their IP
> addresses regardless of the lifetime. Make sense ?

Two things to note here:

1) Since it may well be the case that, even if we specify a mechanism
like that, it may take a while before it is implemented and deployed, it
still makes sense to implement smarts at hosts (even when this doesn't
preclude improvements on the router side).

2) In principle, a host may infer that if, say, two consecutive RAs from
the same router do not contain the previously advertised prefix, the
prefix has become stale -- yes, might require more time for the stale
prefixes to be discarded (the amount of time for 'n' unsolicited RAs to
be transmitted), but can readily be implemented at hosts without the
need to change routers -- not that changing routers would be bad... but
from the host's point of view, you could never tell if the router will
signal prefix changes or not.

Thanks!
-- 
Fernando Gont
e-mail: fernando en gont.com.ar || fgont en si6networks.com
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1