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

[Mikael Abrahamsson <[e-mail]>]

On Thu, 31 Jan 2019, Fernando Gont wrote:

> Question:
> How about making the update to the "Valid Lifetime" a SHOULD or MAY,
> instead?  Or, maybe, at least, set it to the current "Valid Lifetime"?
> Otherwise, with the default values for the "Valid Lifetime", the
> addresses might lie around for 1 month....

I had a real life example today. I have OpenWrt 18.06.1 HGW. This morning, 
I lost Internet access. After diagnosing I realised that WAN was not 
working correctly, so I did "ifdown wan" and "ifdown wan6", switched to my 
secondary provider (which I was still on contract with but don't normally 
use) and then did "ifup" on both. I received new IPv4 GUA and IPv6 GUA PD 
/56, which was then announced on my br-lan. I then ran on that for 3-4 
hours until my regular provider had fixed their hardware problem. Then I 
did the whole ifddown procedure, swapped WAN cable to normal ISP, and did 
ifup again.

Now, I saw IPv6 IA_NA for the temp ISP /56 with a week lifetime on wired 
LAN device so I was a bit worried this would cause long-time problems, but 
the prefixes were deprecated and I couldn't find any actual issues (source 
address selection worked properly). A few hours later all these addresses 
were gone even from my purely wired devices that hadn't seen any link 
down/up events.

So potentially RFC7084 style solves most problems together with 
https://tools.ietf.org/html/draft-patterson-intarea-ipoe-health-05 style 
WAN down/up detection for the IPoE deployment case (PPPoE already have 
this).

OpenWrt since 15.05 has been very much aimed to be RFC7084 compatible so I 
would propose to investigate further what hosts today actually do in 
reaction to RFC7084 signalling.

-- 
Mikael Abrahamsson    email: [e-mail]