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

[Nick Hilliard <[e-mail]>]

Joel M. Halpern wrote on 03/02/2019 18:30:
> Have I missed something in the "problem"?  Robustness is nice.  But 
> let's not bend over backwards trying to fix a problem of multiple failures.

the issue is that when this happens, it causes a cascading failure 
situation which can only be fixed either by the end user manually 
prodding all the ipv6 interfaces on the network to refresh the prefix, 
or if the slaac prefix times out naturally across the home network.

If this doesn't happen, the end user experience is going to be that ipv6 
stops cat videos and I.M. apps from working properly.  After a quick 
call-out, the family tech-support dude will cheerily confirm that this 
can be fixed by disabling ipv6, as she/he fumes while rebooting every 
goddamned device on the network because granny moved the cable modem and 
that jiggled the cable, third time this week (YU NO WORK??!?!?).

Mikael pointed out a related problem that if the cpe is delegated a 
prefix via dhcp6-pd and splits this across multiple home networks, it 
needs to do this in a repeatable way so that even if the SP remembers 
what prefix it previously delegated, the CPE will divvy that up the same 
way it did before the reboot.

For whatever sets of underlying reasons, this is a real world problem, 
which indicates that between CPE vendors and provisioning system 
vendors, there are problems which lead to end user loss of service. The 
IETF may well be able to offer reasonable guidance here.

Nick