Release Asuswrt-Merlin 386.4 is now available

[Lee MacMillan]

Dirty flash from 386.3_2 about 10 hours ago. Everything works as expected on my very simple setup.

 

[explorer200]

It may just be me, but I find logging into AC86U, or AX88U via HTTPS is woefully slow.
Navigating around once in is also pretty awful - very slow.
Via HTTP is snappy & no problems.
I would be happy to use HTTPS, but not the way it is.

(This may not be your issue, just sharing what happens here….)

In addition to my local AX88U I did a remote update to 386.4 via OpenVPN on an AX88U, AX86U, AC86U, and 2 AC68U's (as nodes). It went flawlessly on all routers except now I can't log into the remote AC86U interface but everything on that network seems to be working (remote security cameras, VPN, etc)

I get to the router.asus.com login page, but when I enter my credentials the page doesn't load.

Not worried or complaining, just noting it. I bet it'll work when I get a chance to manually reboot it.

 

[Crimliar]

RT-AC86U dirty flash from 386.3.2 no issues regards performance (inc router menu HTTPS works speedy and fine. There is a new error popping up though:

Jan 5 01:13:44 router dnsmasq-script[4360]: json_object_from_file: error opening file /jffs/nmp_vc_json.js: No such file or directory

 

[ColinTaylor]

RT-AC86U dirty flash from 386.3.2 no issues regards performance (inc router menu HTTPS works speedy and fine. There is a new error popping up though:

Jan 5 01:13:44 router dnsmasq-script[4360]: json_object_from_file: error opening file /jffs/nmp_vc_json.js: No such file or directory

Already noted multiple times, including here.

 

[albinoman887]

Hi,

I also have similar problems after upgrading my GT-AX11000 to Merlin 386.4, while version 386.3_2 runs smoothly without a glitch.

The TP-Link smart bulbs, switches and plugs started to disconnect and re-connect from the 2.4 wifi network randomly after the router being upgraded to version 386.4.
I noticed from the wireless log that the P flag (Powersave mode) goes on and off randomly with all the TP-Link devices, whenever the P flag is ON, that particular device is disconnected.
I fall back to 386.3_2 and check the wireless log, the P flag never comes on and the devices had no connection problems at all.
Anyone has any ideas if there is a router setting that will stop activating the Powersave feature of the View attachment 38226devices?

Just skimmed the whole thread. I have some comments:

something is borked with asus's 2.4ghtz wifi driver. I actually had issues with some of my RGB light strips on the last stable version and not this one. They would reboot about once a day.

Hi,

I also have similar problems after upgrading my GT-AX11000 to Merlin 386.4, while version 386.3_2 runs smoothly without a glitch.

The TP-Link smart bulbs, switches and plugs started to disconnect and re-connect from the 2.4 wifi network randomly after the router being upgraded to version 386.4.
I noticed from the wireless log that the P flag (Powersave mode) goes on and off randomly with all the TP-Link devices, whenever the P flag is ON, that particular device is disconnected.
I fall back to 386.3_2 and check the wireless log, the P flag never comes on and the devices had no connection problems at all.
Anyone has any ideas if there is a router setting that will stop activating the Powersave feature of the View attachment 38226devices?

This powersave toogling issue i also noticed with my problem lights. I eventually just gave up. they only do it once a day or so and its for only a few secs.

Now the TP-Link issues....I had a similar issue on the last beta i installed. I think it is a conflict with DHCP reservations. When i had this issue i noticed in the log something about an its ip address being already in use. What i did to fix it was set up 2 static DHCP leases for each of the problem devices. On the first entry assign the "problem ip" to a new , made up mac address. then set another reservation with a new never used IP and point that to the Mac address of the problem device. reboot everything. If that works. you can remove the "dummy" lease now.

to better describe this i'll make an example

Problem device: TP LINK
MAC: 11 22 33 44 55 66 77 88
IP 10.0.0.5 (check syslog for this if you dont know what it was)

LAN Setup Page > DHCP

Add an entry that is as follows
IP: 10.0.0.5 :MAC AA BB CC DD EE FF
click apply

Now add another entry as follows
IP 10.0.0.6 MAC 11 22 33 44 55 66 77 88

Reboot

If that doesnt work unplug EVERYTHING all switches EVERYTHING for 10 mins

now on to the dns querys...

@RMerlin

I think we can all admit this new behavior is at the very least , not ideal. I was just wondering, why not just rever the code to the last known working revision. Just the code related to network monitoring... I mean thats one of the benefits of being able to compile your own builds, change the little annoyances .. Or is this another one of those roadblocks because of closed source elements ? I My OCD/perfectionist is leaving that nagging thought in the back of mind of that every 30 secs ping and how unnessassary it seems and when you only have DSL, every little bit of bandwith matters...

Another thing

I forgot to quote the user who posted, but the guy having issues with Adaptive QOS. try disabling hw acceleration (runner and flow control)

trun these commands

nvram set fc_disable_force=1
nvram set runner_disable_force=1
nvram commit
service reboot

I dont know the exact cause of the issue but it seems Adaptive QoS in merlin AND IN STOCK has a bug in it that cause a huge performance hit when hw accell is active. Ive tested this extensively in stock and merlin. ANyone can reporduce it pretty easy. do a bufferbloat test with it still enabled first....

Bufferbloat and Internet Speed Test - Waveform

Is bufferbloat causing issues with your internet connection? Run this test to find out.

www.waveform.com

under load i bet my left nut the latency witll increase to like +50ms or more...

turn off runner and fc as stated, reboot, run the same test again.. should get like +5 to +10 at most if you have your bandwith limits set correct. This is why some people swear by cake i think. cake by nature disables fc and runner. Technically you can disable just flow control but idk how that would effect other things since if i understand right, the go hand and hand...

Ok now on to my question.

Has anyone noticed strange behavior with the reporting of your total ram? i did a full reset including usb stick format. when i made my swap partition again, i noticed some new size options. all the way to 5GB so it ried that just for kicks. It completed fine but in sysinfo it showed swap total size as only 1GB and RAM as 512MB. this is on a AX86U. so i went back and remade the swap as 2GB and now it reports swap correct but the real ram is still all wonky... I have a feeling its just a error in the gui reporting it but figured id let everyone know

 

[ColinTaylor]

Has anyone noticed strange behavior with the reporting of your total ram? i did a full reset including usb stick format. when i made my swap partition again, i noticed some new size options. all the way to 5GB so it ried that just for kicks. It completed fine but in sysinfo it showed swap total size as only 1GB and RAM as 512MB. this is on a AX86U. so i went back and remade the swap as 2GB and now it reports swap correct but the real ram is still all wonky... I have a feeling its just a error in the gui reporting it but figured id let everyone know

This should have been fixed in this release. Post your details in this thread:

http://www.snbforums.com/threads/displayed-memory-usage-error-in-webui.76204/post-729399

 

[abracadabra11]

Is anyone experiencing DNS resolution issues with 386.4? I've been having some issues getting webpages to load and I don't know if it's related to the update to 386.4, recent snowfall and down equipment (noticed my local IP address changing a lot in the past two days) or perhaps some other issues with the DNS server (cloudflare). There aren't any obvious issues with dnsmasq that I see other than some bogus replies to queries.

I've tentatively changed my DNS server and will monitor.

Following dnsmasq log didn't show any obvious issues. But some websites simply were inaccessible and couldn't be reached when on 386.4.

Reverted back to 386.3_2 after continuing to run into issues trying to reach websites on 386.4. Websites were immediately reachable after reverting.

 

[shabbs]

something is borked with asus's 2.4ghtz wifi driver.

Yeah, something has been tweaked and it's not playing nice with some devices.

 

[ColinTaylor]

I didn't lose my DHCP hostnames upon my downgrade to 386.3_2. That is assuming these are DHCP hostnames and I'm not confused.
View attachment 38273

That is most likely either a Network Map client name or a host name pushed out by the device itself. DHCP host names are manually assigned under LAN - DHCP Server.

 

[lllSpylll]

I upgraded my RT-AC68u with 2 mesh nodes from 386.3_2 to 386.4 and I have 2 issues:

1) My router shows disconnected every couple of hours. I need to reboot it to fix the issue. I've tried the workaround suggested here but it doesn't fix the problem for me.

2) I cannot create a working guest Wi-Fi on 2.4GHz. The wifi is created and I can connect to it, but it doesn't give internet access. I don't allow my guest Wi-Fi bro access the intranet FYI.

I also use a pihole DNS server if that matters but I don't see any problems in the logs.

 

[slayerking]

Tried one last time last night to go to 386.4 from 386.3_2 and it went horrible again.
Pages like AiProtection and Wan (there are more) not loading properly so can't make any changes, same wifi dropping out with messages of wifi not connected. DNS errors all over the place.
Kids yelling at me because their ping went from 120 to 3000. Will wait for the next release this time as something is hating my AX88U

 

[albinoman887]

so i havent investigated which exact configuration option is the culprit but I have bandwith monitor working without DNS query enabled (as far as i can tell) unless it is hidden from all the regular logging (dmsg syslog proc)

 

[albinoman887]

Tried one last time last night to go to 386.4 from 386.3_2 and it went horrible again.
Pages like AiProtection and Wan (there are more) not loading properly so can't make any changes, same wifi dropping out with messages of wifi not connected. DNS errors all over the place.
Kids yelling at me because their ping went from 120 to 3000. Will wait for the next release this time as something is hating my AX88U

clear your browser cache

 

[Tech9]

I'm giving up testing this release on AC86U due to intermittent 5GHz Wi-Fi issues. It's happening with different Wi-Fi clients, not one only. Nothing in logs, just DHCP message. Stock Asuswrt 45956 Wi-Fi is stable, including AiMesh. It was a fight to add AC68U as node with 386.4, worked first time in Asuswrt. See you at Asuswrt RC3 release and/or next Asuswrt-Merlin.

 

[kernol]

so i havent investigated which exact configuration option is the culprit but I have bandwith monitor working without DNS query enabled (as far as i can tell) unless it is hidden from all the regular logging (dmsg syslog proc)

Have no input on your post - other than to recommend a utility that the Maestro recommended for your screenshots ...
https://greenshot.org/
It works a treat ... use it or lose it - your call.

 

[slayerking]

clear your browser cache

This browser literally only saw the router, Was a clean install of arch and installed browser>opened router>ran update>borked router. I'm with Tech9 I'll be back next release.

 

[albinoman887]

Just skimmed the whole thread. I have some comments:

something is borked with asus's 2.4ghtz wifi driver. I actually had issues with some of my RGB light strips on the last stable version and not this one. They would reboot about once a day.

This powersave toogling issue i also noticed with my problem lights. I eventually just gave up. they only do it once a day or so and its for only a few secs.

Now the TP-Link issues....I had a similar issue on the last beta i installed. I think it is a conflict with DHCP reservations. When i had this issue i noticed in the log something about an its ip address being already in use. What i did to fix it was set up 2 static DHCP leases for each of the problem devices. On the first entry assign the "problem ip" to a new , made up mac address. then set another reservation with a new never used IP and point that to the Mac address of the problem device. reboot everything. If that works. you can remove the "dummy" lease now.

to better describe this i'll make an example

Problem device: TP LINK
MAC: 11 22 33 44 55 66 77 88
IP 10.0.0.5 (check syslog for this if you dont know what it was)

LAN Setup Page > DHCP

Add an entry that is as follows
IP: 10.0.0.5 :MAC AA BB CC DD EE FF
click apply

Now add another entry as follows
IP 10.0.0.6 MAC 11 22 33 44 55 66 77 88

Reboot

If that doesnt work unplug EVERYTHING all switches EVERYTHING for 10 mins

now on to the dns querys...

@RMerlin

I think we can all admit this new behavior is at the very least , not ideal. I was just wondering, why not just rever the code to the last known working revision. Just the code related to network monitoring... I mean thats one of the benefits of being able to compile your own builds, change the little annoyances .. Or is this another one of those roadblocks because of closed source elements ? I My OCD/perfectionist is leaving that nagging thought in the back of mind of that every 30 secs ping and how unnessassary it seems and when you only have DSL, every little bit of bandwith matters...

Another thing

I forgot to quote the user who posted, but the guy having issues with Adaptive QOS. try disabling hw acceleration (runner and flow control)

trun these commands

nvram set fc_disable_force=1
nvram set runner_disable_force=1
nvram commit
service reboot

I dont know the exact cause of the issue but it seems Adaptive QoS in merlin AND IN STOCK has a bug in it that cause a huge performance hit when hw accell is active. Ive tested this extensively in stock and merlin. ANyone can reporduce it pretty easy. do a bufferbloat test with it still enabled first....

Bufferbloat and Internet Speed Test - Waveform

Is bufferbloat causing issues with your internet connection? Run this test to find out.

www.waveform.com

under load i bet my left nut the latency witll increase to like +50ms or more...

turn off runner and fc as stated, reboot, run the same test again.. should get like +5 to +10 at most if you have your bandwith limits set correct. This is why some people swear by cake i think. cake by nature disables fc and runner. Technically you can disable just flow control but idk how that would effect other things since if i understand right, the go hand and hand...

Ok now on to my question.

Has anyone noticed strange behavior with the reporting of your total ram? i did a full reset including usb stick format. when i made my swap partition again, i noticed some new size options. all the way to 5GB so it ried that just for kicks. It completed fine but in sysinfo it showed swap total size as only 1GB and RAM as 512MB. this is on a AX86U. so i went back and remade the swap as 2GB and now it reports swap correct but the real ram is still all wonky... I have a feeling its just a error in the gui reporting it but figured id let everyone know

where it says Total: i think it is actually supposed to say Used...?

 

[Tech9]

where it says Total: i think it is actually supposed to say Used...?

Part of the RAM is used by the system. AC86U has 512MB RAM installed and it shows 430MB total in Tools.

 

[BobbyFlobby]

I had issues with the alphas and betas, wifi dropping off, devices not connecting on both 2.4 and 5ghz and cams also dropping etc.
running aimesh dual AX86U via lan cable.
so i decided to bite the bullet and nuke them both and clear nvram, and so far for the last 2 days, everything is working normally as per the last build

Getting a lot of the following notices in the logs (is it due to 5ghz auto channel?), but everything seems to be working so its not a problem, will lock the channel down and see if it clears it.

acsd: acs_candidate_score_intf(1133): eth7: intf check failed for chanspec: 0xe832 (36/160)
acsd: acs_candidate_score_bgnoise(1602): eth7: bgnoise check failed for chanspec: 0xe832 (36/160)
acsd: acs_candidate_score_txop(1860): eth7: txop check failed for chanspec: 0xe832 (36/160)

 

[joomlafab]

Dirty upgrade from 386.3_2 on AX3000 worked fine. No problem at all after running 48 hours. Many thanks to RMerlin and everyone involved.