RB5009 IPv6 prefer bridge as WAN instead of pppoe-out1

I have several IPIPv6 Tunnel to connect CGNAT device, all of sudden IPv6 Routing prefer IPv6 Address from “bridge” interface, not proper IPv6 WAN from “pppoe-out”

This because Telekom Malaysia stupid enough to use RA instead of DHCPv6 for PPPoE:
Screenshot 2025-02-19 223548.png
Screenshot 2025-02-19 223150.png
IPv6 Address (truncated):

Flags: D - DYNAMIC; G - GLOBAL, L - LINK-LOCAL
Columns: ADDRESS, FROM-POOL, INTERFACE, ADVERTISE, VALID
 #    ADDRESS                      FROM-POOL   INTERFACE   ADVERTISE  VALID
 3  L fe80::beef/64                            pppoe-out1  no
 5  G 2001:e68:5450:414::/64       pppoe-out1  bridge      yes
 6  L fe80::1/64                               bridge      no
11 DL fe80::f91a:be7c:0:ab/64                  pppoe-out1  no
12 DG 2001:e68:5090:cdf2::beef/64              pppoe-out1  no         2d23h45m43s

IP > Cloud:

[admin@MikroTik] > /ip cloud print
          ddns-enabled: yes
  ddns-update-interval: 1m
           update-time: yes
        public-address: 175.138.216.213
   public-address-ipv6: 2001:e68:5450:414::
              dns-name: xxxxxxxxxxxx.sn.mynetname.net
                status: updated
      back-to-home-vpn: revoked-and-disabled

the “public-address-ipv6: 2001:e68:5450:414::” is unusable for IPIPv6 Tunnel, or any service want to connect this RB5009

Temporary fix is I remove all ::/0 in IPv6 Routes, then I add manually like this:
Screenshot 2025-02-19 224555.png
Update Cloud DDNS, now using correct IPv6 WAN from PPPoE that given by ISP RA:

[admin@MikroTik] > /ip cloud print
          ddns-enabled: yes
  ddns-update-interval: 1m
           update-time: yes
        public-address: 175.138.216.213
   public-address-ipv6: 2001:e68:5090:cdf2::beef
              dns-name: xxxxxxxxxxxx.sn.mynetname.net
                status: updated
      back-to-home-vpn: revoked-and-disabled

This only temporary, every 3 days, I must delete dynamic routes, this behavior exist ROS 7.18beta6, 7.17.2 and below

but, every time I delete IPv6 dynamic routes to fix, cause this issue on ROS 7.18beta6:

2025-02-18 23:27:08 system,clock,critical,info ntp change time Feb/18/2025 23:26:12 => Feb/18/2025 23:27:08
2025-02-18 23:27:11 system,error,critical router was rebooted without proper shutdown, probably kernel failure
2025-02-18 23:27:11 system,error,critical kernel failure in previous boot
2025-02-18 23:27:48 system,clock,critical,info ntp change time Feb/18/2025 23:27:48 => Feb/19/2025 21:18:31
2025-02-19 21:18:35 system,error,critical router was rebooted without proper shutdown, probably kernel failure
2025-02-19 21:18:36 system,error,critical kernel failure in previous boot

I think I found another solution, by delete ISP default v6 route leaving route to PPPoE AC and RB5009 use pppoe-out1 address instead of bridge address

Infinite loop to delete ISP route:
Screenshot 2025-03-01 140501.png
remove ::/0 gateway pppoe-out1
Screenshot 2025-03-01 140529.png
usually Telekom Malaysia give RA lease for 3 days, removing Route given by ISP cause glitch and I stay same address indefinitely
Screenshot 2025-03-01 140548.png
* Valid & Preferred time increasing instead of decrease

Cloud DDNS now using correct pppoe-out1 address!
Screenshot 2025-03-01 140608.png