Release Asuswrt-Merlin 3006.102.7 is now available

[dinko05]

Unfortunately, for my BE92U, I still got a kernel crash with the time issue on the latest pre-release with the newest GPL. Here is the latest crash:

```
Mar 15 10:22:57 et12 hostapd: wl2.2: STA 00:00:00:00:00:00 IEEE 802.11: disassociated
Aug 24 04:56:16 router kernel: rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
Aug 24 04:56:16 router kernel: rcu: 0-...!: (2 ticks this GP) idle=dc6/1/0x4000000000000000 softirq=7290684/7290684 fqs=0
Aug 24 04:56:16 router kernel: rcu: 1-...!: (222 ticks this GP) idle=2ba/1/0x4000000000000000 softirq=5063391/5064327 fqs=0
Aug 24 04:56:16 router kernel: rcu: (detected by 2, t=900726557960 jiffies, g=3071113, q=45)
Aug 24 04:56:16 router kernel: tdts_core_ioctl_udb_op_prog_ctrl() fail!
Aug 24 04:56:16 router kernel: rcu: rcu_sched kthread starved for 900726557960 jiffies! g3071113 f0x2 RCU_GP_WAIT_FQS(5) ->state=0x200 ->cpu=2
Aug 24 04:56:16 router kernel: rcu: RCU grace-period kthread stack dump:
Sep 29 11:24:32 router chronyd[12818]: Forward time jump detected!
Aug 24 04:56:16 router kernel: [BLOCKED - INBOUND] IN=vlan4094 OUT= MAC=60:cf:84:3a:a9:a0:40:48:6e:bc:b6:31:08:00:45:00:00:2c SRC=193.163.125.7 DST=172.1.136.40 LEN=44 TOS=0x00 PREC=0x00 TTL=242 ID=40195 PROTO=TCP SPT=33207 DPT=2762 SEQ=1412969829 ACK=0 WINDOW=14600 RES=0x00 SYN URGP=0 OPT (020405B4)
Aug 24 04:56:16 router kernel: rcu_sched R 0 11 2 0x00000008
Aug 24 04:56:16 router kernel: Call trace:
Aug 24 04:56:16 router kernel: __switch_to+0x148/0x170
Aug 24 04:56:16 router kernel: [BLOCKED - INBOUND] IN=vlan4094 OUT= MAC=60:cf:84:3a:a9:a0:40:48:6e:bc:b6:31:08:00:45:00:00:34 SRC=85.217.149.69 DST=172.1.136.40 LEN=52 TOS=0x00 PREC=0x00 TTL=46 ID=44471 PROTO=TCP SPT=41441 DPT=5858 SEQ=4017440342 ACK=0 WINDOW=65535 RES=0x00 SYN URGP=0 OPT (020405B40103030801010402)
Aug 24 04:56:16 router kernel: __schedule+0x214/0x5a0
Aug 24 04:56:16 router kernel: schedule+0x38/0xa0
Aug 24 04:56:16 router kernel: schedule_timeout+0x15c/0x2a0
Aug 24 04:56:16 router kernel: rcu_gp_kthread+0x488/0x900
Aug 24 04:56:16 router kernel: [BLOCKED - INBOUND] IN=vlan4094 OUT= MAC=60:cf:84:3a:a9:a0:40:48:6e:bc:b6:31:08:00:45:00:00:28 SRC=185.224.128.16 DST=172.1.136.40 LEN=40 TOS=0x00 PREC=0x00 TTL=242 ID=54321 PROTO=TCP SPT=33144 DPT=9000 SEQ=101545099 ACK=0 WINDOW=65535 RES=0x00 SYN URGP=0
Aug 24 04:56:16 router kernel: kthread+0x118/0x150
Aug 24 04:56:16 router kernel: ret_from_fork+0x10/0x24
Aug 24 04:56:16 router crond[4808]: time disparity of 15012122 minutes detected
```

It still seems more stable than 102.7_beta2 but unfortunately, the crash is still there at least for me (note the et12 line is coming from the et12 mesh node since all logs seem re-routed to my external log server; it's just to give a sense of the time jump).

Thanks,
Romain

 

[RMerlin]

Unfortunately, for my BE92U, I still got a kernel crash with the time issue on the latest pre-release with the newest GPL. Here is the latest crash:

Double check that you are indeed running the latest test build. The version at the top should report "3006.102.7_1-gbeea6c005a"

Mine still hasn't recrashed since the update, while previously it would crash within a few minutes after booting.

 

[dinko05]

On reboot it listed this: ```Dec 31 16:01:23 router init: fwver: 3.0.0.6_102.7_1-gbeea6c005a``` which seems to match what you mentioned. I have reinstalled it again to make sure it was all properly installed and will let you know.

For me, it was never a "crash soon after reboot" but more "crash randomly". I do have a lot of things running on it but nothing super exotic.

 

[Samuel_1975]

Works pretty ok for me and my dad with rt-be92u so the test version seems to work

 

[Igor]

I am not seeing an issue with trying to add a fourth mesh node.

Thanks, so something's wrong with me.
Are there any ways to figure out the cause of the problem?

 

[Zirescu]

BE92U test build 3006.102.7_1-g4fd221f466 issues:
- WAN DHCP didn't renew and had to force a reboot
- Network Map -> View/Client list -> Network tab is showing my Wired devices under the MAIN NETWORK SSID associated to the 5/6GHz network. They do show up correctly under the Network client tab view.

 

[greig2000uk]

many thanks for that, i'll away and update now

many thanks for that, i'll away and update now

Shouldnt have bothered, nothing works now!
No Ai apps work outside my network, is there something I am missing here, a setting needing done or something i've missed

 

[argh0815]

Shouldnt have bothered, nothing works now!
No Ai apps work outside my network, is there something I am missing here, a setting needing done or something i've missed

What are the server settings in your AiFoto app? Are you using a DDNS service, which port, http or https? The port you're using must be open on your router, either forward it manually to your NAS or you could potentially enable UPNP.

 

[random9]

Potential Bug Report:
Model: RT-BE96U
Dirty upgrade from .6 to .7 (and since a dirty downgrade back to .6)

1) .7 seems to have problems with DHCP from my modem configured in bridge mode. Rebooting the modem after the router allowed the router to obtain an IP. However, additional reboots of the router resulted in no IP until rebooting the modem again. After the dirty downgrade back to .6, it dhcp's from the modem perfectly fine. Modem: Xfinity XB8

2) Media (WiFi) bridges failed to connect to .7. I have 2- one is an old Asus router and the other is Netgear. Both worked with older firmware versions and both are failing to connect with .7. Unfortunately, the dirty downgrade back to .6 does not seem to have resolved this. Not sure what to do here.

 

[MDM]

Potential Bug Report:
Model: RT-BE96U
Dirty upgrade from .6 to .7 (and since a dirty downgrade back to .6)

1) .7 seems to have problems with DHCP from my modem configured in bridge mode. Rebooting the modem after the router allowed the router to obtain an IP. However, additional reboots of the router resulted in no IP until rebooting the modem again. After the dirty downgrade back to .6, it dhcp's from the modem perfectly fine. Modem: Xfinity XB8

2) Media (WiFi) bridges failed to connect to .7. I have 2- one is an old Asus router and the other is Netgear. Both worked with older firmware versions and both are failing to connect with .7. Unfortunately, the dirty downgrade back to .6 does not seem to have resolved this. Not sure what to do here.

Simple - Dirty upgrade (or downgrade) is not the way to test anything...
Alas as you fear - only factory defaults is.

 

[random9]

Simple - Dirty upgrade (or downgrade) is not the way to test anything...
Alas as you fear - only factory defaults is.

Sure, but I was not trying to test. I was hoping to do an in-place upgrade from a final/release version to the next final/release version. Of all the various features and functions that could fail without a factory default, I'm pretty surprised with those two that I reported.

 

[RMerlin]

1) .7 seems to have problems with DHCP from my modem configured in bridge mode. Rebooting the modem after the router allowed the router to obtain an IP. However, additional reboots of the router resulted in no IP until rebooting the modem again. After the dirty downgrade back to .6, it dhcp's from the modem perfectly fine. Modem: Xfinity XB8

Nothing was changed with regards to the WAN handling between 102.6 and 102.7.

merlin@ubuntu-dev:~/dev/amng$ git log --oneline 3006.102.6..3006.102.7
21d30a367f (tag: 3006.102.7) Bumped revision to 3006.102.7 final
7b6810ec20 Updated documentation
2bc9ee0923 rc: refresh wireguard BLOG bypass when refreshing VPN routing on BCM4916 platform
b4182dd67a webui: update routing rules when toggling a client from the VPNDirector page
4904a33e94 webui: styling fixes to the Firmware Upgrade page for non-AiMesh models.
0099cd1cb9 Bumped revision to beta 2
96e757e5f9 Updated documentation
3f2c98fe2f rom: fix missing signature updater for models with TrendMicro HNS engine
ec8eeec434 (tag: 3006.102.7-beta1) Bumped revision to beta 1
f17e8fd1b0 dnsmasq: Fix PXE boot server (PXEBS) responses broken in 2.92
1cb558f26e VSCode: filter out parsing of the multiple kernels + SDKs
371168d9f7 openvpn: default server keysize to 2048 bit
c8043ed06b Updated documentation
78f6248cdf tor: updated to 0.4.8.22
38c5c23e79 openvpn: updated to 2.6.19
d852901046 Updated documentation
11a4dfd13f webui: display ntpd options in WISP mode
cda1c30121 Bumped revision to alpha 2
ee6ec0e3c8 webui: stop polling online copy of DNS DB on WAN page
a855f2bb40 webui: disable obsolete DNS DB pulling on WAN page
64f6a2a0ee Updated documentation
0fe770bb67 rc: Add new 2024 IANA Root trust anchor
ddfdb4e847 webui: fix Connections table layout (wrong class)
fcc481062c webui: remove duplicate code on WAN page
5eb90360bb webui: add missing UPnP settings to multiservice version of WAN page
d744657e68 openssl: Ported 7 OpenSSL-1.1.1w CVE Fixes from Ubuntu ESM Repository. (#884)
e0071bcad0 dnsmasq: updated to 2.93-test2
5b063cb580 Updated documentation
77204f70e6 Updated documentation
56a7cf7003 dnsmasq: updated to 2.92
c30f5681c8 Updated documentation
091ab74b7a dropbear: update to 2025.89
92bcc4a553 nettle: updated to 3.10.2
e9933de16e openvpn: update to 2.6.17
944609be6d aicloud: disabled AiCloud
47316c04f3 Bumped revision to 3006.102.7 alpha 1
4090acdc79 rc: fix missing DNS Director DoT IPv6 rules (#883)

 

[greig2000uk]

What are the server settings in your AiFoto app? Are you using a DDNS service, which port, http or https? The port you're using must be open on your router, either forward it manually to your NAS or you could potentially enable UPNP.

same as they were before - HTTPS, when all the Ai apps worked on 3006.102.5 - nothing changed
update to .7 nothing works, back to .5, nothing works
There is most likely some setting I need to make but I just cant figure it out

help appreciated

 

[argh0815]

same as they were before - HTTPS, when all the Ai apps worked on 3006.102.5 - nothing changed
update to .7 nothing works, back to .5, nothing works
There is most likely some setting I need to make but I just cant figure it out

help appreciated

Is UPNP enabled on your router (Advanced Settings/WAN)? With 3006.102.6, the default setting was changed to disabled.

 

[greig2000uk]

Is UPNP enabled on your router (Advanced Settings/WAN)? With 3006.102.6, the default setting was changed to disabled.

I never used .6 and already went from .7 back to .5 - currently back on .7 so it will be disabled but going back to .5 yesterday I still had the problem
I'll try both again tonight

 

[greig2000uk]

I never used .6 and already went from .7 back to .5 - currently back on .7 so it will be disabled but going back to .5 yesterday I still had the problem
I'll try both again tonight

so upnp is enabled - I set up open vpn but its a pain in the arse to backup pictures using it so not prefered.
Now I have enabled EZ-Connect on my NAS and all is working as it should though I'm not too happy having to use this method.

would still like some help on this, I dont understand why going back to .5 didn't put everything back as it was?

 

[Saulus]

Subject: RT-AX86U Pro (3006.102.7): Dual WAN Load Balance broken & Failover requires reboot

Message:Hi RMerlin and community,

I'm reporting a series of issues regarding Dual WAN functionality on the RT-AX86U Pro running the latest 3006.102.7 firmware. It seems like the routing logic in the 3006 branch has some significant regressions.

My Setup:

• Primary WAN: 2.5G Port (ISP 1).
• Secondary WAN: Ethernet LAN - LAN Port 1 (ISP 2).

Issue 1: Load Balance is completely broken

• When switching to Load Balance mode (even at 1:1), the internet connection drops immediately for all devices.
• Routing Rules issues: If "Enable Routing Rules" is set to "No", there is zero connectivity.
• UI Validation: The UI is extremely strict. It doesn't allow the "Destination IP" field to be empty. Trying to use 0.0.0.0/0 results in an "Invalid IP address" error.
• 'all' keyword bug: While it accepts the keyword all for a specific source IP, it refuses to add a second rule with the same all destination for another WAN unit, saying "This entry was already in the list."

Issue 2: Failover / WAN Switch requires a full reboot

• This is the most frustrating part. Even in Failover mode, if I manually change the "Primary WAN" in the settings to switch my external IP:
  
  • The change does not apply dynamically (even after waiting several minutes).
  • The router requires a full reboot to actually start routing traffic through the newly selected Primary WAN.
• On older firmware versions (pre-3006), this switch used to take only 10-20 seconds without a reboot.
• It feels like the routing table or conntrack is not being properly flushed/updated when WAN settings change in this build.

Troubleshooting done:

• Verified with different cables.
• Checked System Log (shows restart_wireless and cfg_server triggers, but no actual WAN routing update).

Has anyone else noticed this behavior on the 3006 branch? Any suggestions on how to force a routing table flush via SSH without rebooting every time would be greatly appreciated.

 

[octopus]

Тема: RT-AX86U Pro (3006.102.7): Нарушение балансировки нагрузки в двух WAN-сетях, для переключения на резервный сервер требуется перезагрузка.

Сообщение: Привет, RMerlin и сообщество!

Сообщаю о ряде проблем, связанных с работой функции Dual WAN на маршрутизаторе RT-AX86U Pro с последней версией прошивки 3006.102.7 . Похоже, что в логике маршрутизации в ветке 3006 обнаружены существенные регрессии.

Моя конфигурация:

• Основной WAN-порт: 2,5 Гбит/с (провайдер 1).
• Вторичная глобальная сеть (WAN): Ethernet LAN - порт LAN 1 (ISP 2).

Проблема 1: Полная неработоспособность балансировки нагрузки.

• При переключении в режим балансировки нагрузки (даже при соотношении 1:1) интернет-соединение немедленно обрывается на всех устройствах.
• Проблемы с правилами маршрутизации: если параметр "Включить правила маршрутизации" установлен на "Нет", то связь отсутствует.
• Проверка пользовательского интерфейса: Пользовательский интерфейс чрезвычайно строг. Поле «IP-адрес назначения» не может быть пустым. Попытка использовать 0.0.0.0/0 приводит к ошибке «Неверный IP-адрес».
• Ошибка с ключевым словом 'all': хотя система принимает ключевое слово all для конкретного IP-адреса источника, она отказывается добавить второе правило с тем же адресом назначения all для другого WAN-устройства, выдавая сообщение "Эта запись уже была в списке".

Проблема 2: Для переключения на резервный канал/WAN-коммутатор требуется полная перезагрузка.

• Это самая неприятная часть. Даже в режиме отказоустойчивости, если я вручную меняю "Основной WAN" в настройках, чтобы переключить свой внешний IP-адрес:
  • Изменения не применяются динамически (даже после ожидания в течение нескольких минут).
  • Для начала маршрутизации трафика через выбранный основной WAN-порт маршрутизатору требуется полная перезагрузка .
• В более старых версиях прошивки (до 3006) этому коммутатору требовалось всего 10-20 секунд без перезагрузки.
• Создается впечатление, что в этой сборке таблица маршрутизации или conntrack некорректно очищаются/обновляются при изменении настроек WAN.

Устранение неполадок выполнено:

• Проверено с использованием разных кабелей.
• Проверил системный журнал (показывает срабатывания restart_wireless и cfg_server, но фактического обновления маршрутизации WAN нет).

Кто-нибудь ещё замечал подобное поведение в ветке 3006? Буду очень благодарен за любые предложения о том, как принудительно очистить таблицу маршрутизации через SSH без перезагрузки каждый раз.

In english pls!

 

[Saulus]

In english pls!

Sorry

 

[penguin22]

Several days ago, I mentioned a new firmware that was released for the BE86U, which was causing me constant problems. It has since been completely stable, going on a week without a single client drop, including multiple IoT networks. I noticed today that the same update language was added to the BE88U that was just released: RT-BE88U｜WiFi Routers｜ASUS USA.

Hopefully this is an indicator of an updated GPL coming to Merlin in the near future, though I wouldn't be surprised if some of these are already included since the GPL releases don't always line up.