Hello, Router/Firewall host is running FreeBSD 14-STABLE: FreeBSD 14.2-STABLE #20 n270632-859aa726fb86: Fri Feb 28 19:38:05 CET 2025
I'm using mpd5(8) to connect to our ISP via vDSL. Utilizing an appropriate "link-up.sh" script, which effectively does - restart rtsol on tun0 (rtsol tun0 &) - restart dhcp6c (service dhcp6 restrt) - doing some logging - performing some DDNS adjustments with the appropriate provider mpd5 is configured to obtain IPv4 and IPv6 via ipcp, ipv6cp. While IPv4 has never been a problem, it seems that IPv6 is stuck with SLAAC (I never managed to obtain an IPv6 via DHCP (dhcp6c(8) from ports), always EUI64, privacy mode set). Restarting mpd5 provides only ONE valid IPv6 address on tun0. When ISP is resetting the address assignment usually after 24 hours for both IPv4 and IPv6, I end up very often having at least two or even more, still valid IPv6 addresses (meaning: none of the former assigned IPv6 addresses is marked deprecated or invalid). This renders DDNS useless, since I have no plan how to figure out the valid address. This problem occured recently, I do not know what causes it, I guess it came with a recent STABLE upgrade. How can mpd5 be forced to deprecate an address before obtaining a new one? How to finde out which of the assigned IPv6 addresses is the "old" one and mark it deprecated? I run a simple script searching for "tentative, deprecate and so on" addresses to leave the good one(s) when providing my DDNS provider with the mutually correct IPv6 address of mine. Utilising link-down.sh of mpd5(8) seems a good place to eradicate IPv6 addresses (by filtering out fe80:: or mutually assigned ULA, leaving the valid IPv6 for deletion), but this seems non-conformal to me. A bug or a "feature"? Thanks in advance, Oliver -- A FreeBSD user
pgprWZQPei12n.pgp
Description: OpenPGP digital signature
