ipv6PrefixCheck — Failure to trigger update after router restart

[fiwswe]

I have seen two situations were a restart of the router (FRITZ!Box) did not lead to the updated IPv6 prefix being noticed.

• After a power outage (affecting the router but not the host running the script).
• After a reboot of the FRITZ!Box (due to a software update).

Effectively both cases are similar in that the router suddenly disappears and when it reappears it reconnects to the Internet receiving new IPs and IPv6 prefix. On OpenBSD this will lead to two public IPv6 addresses being present, one with the old IPv6 prefix and another with the new one. The old IPv6 address will be non-functional of course as it will not be routed to the WAN interface of our router. The timeouts before this situation resolves itself can be on the order of 2 hours in my setup.

This situation may cause the script to fail to recognise that the actual IPv6 prefix has changed, until the lifetime of the old public IPv6 address has been reached and the address is removed from the interface.

[fiwswe]

Determining the current public IPv6 prefix may be accomplished more easily by asking slaacd(8) for the advertised prefix using slaacctl(8) instead of parsing routing tables:

slaacctl show interface 'em0'|grep 'prefix:'|grep -v 'prefix: fd'

(Get current slaacd(8) data for interface em0, extract the lines containing the word prefix:, then filter out those that contain the ULA prefix. Extracting the actual prefix is then fairly easy.)

This yields output similar to this:

		prefix: 2001:dead:cafe::/64

This method needs to be explored further to determine if it consistently works …

From there it would be very easy to determine the public IPv6 address of the host on the given interface.