Beta Asuswrt-Merlin 386.4 beta is now available

[gerardr]

Make sure you don't interfere with the router trying to ping a Microsoft server to determine your connection state.

Hi! I'm not pinging anything, that I know of. Both dns_probe and wandog are disabled. Here's my current nvram output:

ax: nvram show | egrep 'dns_probe'\|'wandog'
dns_probe=0
dns_probe_content=
dns_probe_host=
wandog_delay=90
wandog_enable=0
wandog_fb_count=4
wandog_interval=90
wandog_maxfail=90
wandog_target=

Again, the network map first shows a working WAN link, then after 'Verifying' briefly displays shows a Disconnected result like the image @frequenzy showed earlier. [Yes, I was experimenting with wandog parameters to try to debug; originally delay was 0, interval was 10 and maxfail was 12.]

This first showed up for me in 386.4_alpha3. Alpha2 and earlier releases were fine.

 

[ColinTaylor]

This bug is still present in this beta whereby using the Tunnelbroker DDNS client removes changes made by firewall-start at bootup.

[BUG] Tunnelbroker DDNS erases changes made by firewall-start

Router: RT-AX86U Firmware: 386.3_2 Stopping or restarting the Tunnelbroker DDNS client removes any custom changes made by firewall-start. This can happen for example when the WAN interface is bounced or in some circumstances at boot. AFAICT the relevant piece of code is here.

www.snbforums.com

 

[RMerlin]

Hi! I'm not pinging anything, that I know of. Both dns_probe and wandog are disabled. Here's my current nvram output:

ax: nvram show | egrep 'dns_probe'\|'wandog'
dns_probe=0
dns_probe_content=
dns_probe_host=
wandog_delay=90
wandog_enable=0
wandog_fb_count=4
wandog_interval=90
wandog_maxfail=90
wandog_target=

Again, the network map first shows a working WAN link, then after 'Verifying' briefly displays shows a Disconnected result like the image @frequenzy showed earlier. [Yes, I was experimenting with wandog parameters to try to debug; originally delay was 0, interval was 10 and maxfail was 12.]

This first showed up for me in 386.4_alpha3. Alpha2 and earlier releases were fine.

It started with 386.4 alpha because that's when Asus changed the way monitoring works. A lot of people are having detection failure in recent Asus stock firmware as well because they had their firewall configured to block access to dns.msftncsi.com. Make sure that's not the case.

Your router seems to be missing the dns_probe_host variable. Set dns_probe_host to dns.msftncsi.com, which is the default value.

 

[SheikhSheikha]

Dirty upgrade from Alpha3, router is stable and running smoothly

 

[5stringdeath]

My DDNS (google) has a big yellow exclamation point (!) as if there is an error, however it is working just fine and my internal routing resolves fine.

 

[gerardr]

It started with 386.4 alpha because that's when Asus changed the way monitoring works. A lot of people are having detection failure in recent Asus stock firmware as well because they had their firewall configured to block access to dns.msftncsi.com. Make sure that's not the case.

Your router seems to be missing the dns_probe_host variable. Set dns_probe_host to dns.msftncsi.com, which is the default value.

OK, thanks. I've developed a work-around. Since 'stubby' (for DNSoverTLS), can use the system hosts file, and I append my hosts lists to it via my usb_mount script, I set dns_probe_host to something I defined and dns_probe_content to its matching IP address. Now network map keeps my active WAN link data, and my WAN link LED is back to white. Plus I'm fairly sure any traffic stays within the router. It uses a bit of processor power, unfortunately, but no needless network traffic. Given that dns_probe_content values were required (found in defaults.c) and it matches the nslookup of dns.msftncsi.com values from one of my remote servers, I'm not sure why a firewall would come into play at all.

 

[ChatmanR]

Dirty upgrade from Alpha3, router is stable and running smoothly on FiOS gigabit and FlexQOS

---

 

[MNoisy]

Does anyone workaround for openvpn to work with IPV6 enabled yet, or just going back to last alpha still?

Beta 1 resolved the issue. I'm still seeing the errors for "not mesh client, can't delete it", but merlin said to disregard those, so all good so far with the first beta.

Thanks for your hard work @RMerlin

 

[nzwayne]

All's great with Beta1 on my routers for myself & family. Bummer that AXE11000 support delayed, as we await Broadcom to assist in the Rmerlin party.

 

[frequenzy]

Try to turn "Internet connection" button ON/OFF

Just tested and this doesn’t work, the only way to bring it up is to configure the dns probe again

 

[GWater]

No extra script needed, working fine here. I can access both the WAN and any of my LAN devices through the OpenVPN connection, on Android or laptops from outside my LAN.
This has been the case for a least the last couple of years that I have been using this function, no helper scripts needed.
Maybe reset your OpenVPN server settings to defaults, and reconfigure from scratch.

Hi jsbeddow,

Would you mind sharing how you've setup your OpenVPN Server? I've cleared the JFFS partition and done a full factory reset of the router. Apart from setting up an internet connect (PPPoE) and the OpenVPN Server I've made no other configuration changes or installed any add-ons. Still no dice with being able to route to LAN clients using OpenVPN Connect on my android phone. Internet access is fine via VPN.
My current settings below (Advertise DNS and IP range 10.13.13.0 are my only changes).

 

[RMerlin]

Hi, I have recently upgraded to the firmware 386.4_beta1 on my RT-AC88U and immediatly noticed that both download and upload graph almost stopped working and only register a couple of KB on the FlexQoS (The only mod I'm using) tab and nothing at all on Classification tab (Even with FlexQoS uninstalled), I think the entire thing just stopped working. I've fully uninstalled FlexQoS, done a factory restore and the bug persists.

Adaptive QoS seems to be broken on the RT-AC88U. I will have to forward the issue to Asus...

 

[pattiri]

Adaptive QoS seems to be broken on the RT-AC88U. I will have to forward the issue to Asus...

I would say only Classification tab is broken QoS is working fine because, yesterday I've downloaded ~300 GB over steam and torrenting and my wife was using streaming services during this and she didn't have any complaints. Earlier, while I was downloading she used to disconnect even her mobile phone from wifi because of struggling even with web browsing.

 

[RMerlin]

I would say only Classification tab is broken QoS is working fine

It's not. All traffic gets allocated to the primary queue, all the sub-queues get zero traffic. That's why the charts stay empty. They started getting traffic once again when I tested with Trend Micro components extracted from a 386_45987 firmware.

I'll check with Asus next week whether I need updated components for other models, tho they seem to properly allocate traffic to the appropriate classes on at least the SDK6 and HND5.02axhnd platforms (I didn't test the other ones).

 

[tasmansaroha]

Adaptive QoS seems to be broken on the RT-AC88U. I will have to forward the issue to Asus...

I recompile Merlin 386.4 beta2 by using bwdpi_source from GPL_RT-AC88U_3.0.0.4.386.45958-gb20a9e9.tgz. result AiProtection, Parental Controls and Adaptive QoS works

 

[RMerlin]

I recompile Merlin 386.4 beta2 by using bwdpi_source from GPL_RT-AC88U_3.0.0.4.386.45958-gb20a9e9.tgz. result AiProtection, Parental Controls and Adaptive QoS works

It's possible these two files that were different from 45987 were incorrect in my archive (or the AC3100/AC5300 archives contained incorrect versions).

EDIT: nope, all three models had the same version, which is different from the one I extracted from 45987

merlin@ubuntu-dev:~$ md5sum asuswrt.386-45958-ac88/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
d2f1650dac3c246298adb54331f06399  asuswrt.386-45958-ac88/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
merlin@ubuntu-dev:~$ md5sum asuswrt.386-45958-ac3100/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
d2f1650dac3c246298adb54331f06399  asuswrt.386-45958-ac3100/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
merlin@ubuntu-dev:~$ md5sum asuswrt.386-45958-ac5300/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
d2f1650dac3c246298adb54331f06399  asuswrt.386-45958-ac5300/release/src/router/bwdpi_source/prebuild/tdts_udbfw.ko
merlin@ubuntu-dev:~$ md5sum amng/release/src/router/bwdpi_source/prebuild/RT-AC88U/tdts_udbfw.ko
7a10949aedb1d04ffc6fd9b46f7f615c  amng/release/src/router/bwdpi_source/prebuild/RT-AC88U/tdts_udbfw.ko

 

[RMerlin]

DIffing the GPL download from their website against mine I do see one file that's different, so that could be why.

 

[albinoman887]

I recompile Merlin 386.4 beta2 by using bwdpi_source from GPL_RT-AC88U_3.0.0.4.386.45958-gb20a9e9.tgz. result AiProtection, Parental Controls and Adaptive QoS works

huh, i always thought we oculdnt build with AIProtection ourselfs just RMlerlin . i would of been compil;ing my own builds long ago lol

 

[John Adler]

dirty flash over 386.3-2
all working fine so far
thks Merlin!

 

[donat76]

First off, thank you. I really appreciate your hard work. I have the AX88U. The latest release firmware worked fine with IPv6 and Adaptive QOS. With the beta, I'm no longer getting IPv6 addresses to my devices and I'm showing a segmentation fault in DCD (Trend micro) module. Same error message and log entries as someone else in this thread.

On a side note, I have a really obscure problem using Adaptive QoS on the last release firmware. "Protocol is buggy" errors for IPv6 and IPv4. It happens when ever the Trend Micro module is loaded either temporarily by clicking on any of the Adaptive QoS tabs in the GUI or when Adaptive QOS is activated. SSH ing into the router and turning off "runner" and flowcache gets rid of the error. I spent hours figuring out what it was and exactly how to trigger it. I'm not sure much can be done about it except passing on the info to Asus or Trend Micro.

Trend Micro seems to be a common cause of issues. It would be nice to be able to do a stack trace with a debug build of the their binary blob but I'm guessing everything is obfuscated and it's hard to track things down. If this is possible, please let me know.

I'll do a very detailed "bug report" for the "protocol is buggy" issue if for no other reason but to document it for other people if it's not possible to fix it and turn back on hardware acceleration.