mwan3 and IPv6 -> mwan3 use in 25.12.x fixes

[Simon Iremonger (openwrt-devel)]

>> I can say that -- ddns-scripts in 25.12-SNAPSHOT r32607  (with a recent
>> fix included already)  now works fine for IPv6 address detection on
>> interface, network, and URL using checkipv6.dyndns.com  etc...
>> It obviously has been improved since 24.10.5 where many of these did

> I disagree regarding the mode "network". It uses network_get_ipaddr6
> which is unstable - if it returns your public IP after boot, it can
> stop doing that after the first address change because odhcp6c always
> puts new IPs at the end of the list, netifd keeps it like that and
> the helper function returns the first IP.



 > No issues with fw4. Depending on your needs, you could miss certain
 > features in fw4 that are helpful in combination with mwan3.
 > The issue with network_get_ipaddr6 returning the wrong address could
 > occur.


Can these limitations and the current state of affairs be put into :-

https://openwrt.org/docs/guide-user/network/wan/multiwan/mwan3

?  ...



> This is solved in my OpenWrt fork. Patches were submitted, but not
> reviewed or accepted yet. And the patches are not new. That's why
> I started my fork.

Has this reached a stable point, can we look at getting this backported
into master and potentially 25.12 series?  What would be pragmatic and
appropriate?  Would refactored version have a new name, mwan4 etc?

**NB** One classic limitation of mwan3 package, is that it will tend
to disable IPv6 by default.  The 'defaults' definitely need changing
so mwan3 installation does not cause a regression if not yet configured.

This may or may not have been sorted already...  In any case, what would
now be pragmatic and sensible, to document and tidy up the situation for
mwan3 in 25.12 "as-is"?


>> If this needs some testing I may be able to help.  I could
>> certainly see if IPv6-tracking works where it didn't before.
> IPv6 tracking is the simple part and already worked when I started
> using mwan3.

Noted!.


--Simon