Router trouble RT-AX88U

[ciprian.trofin]

Not tested before. The reason is IoT - my Roborock vaccum only connects to 1 SSID. Moving it from one area to another would require a new wireless setup (PITA).

Another thing I did today: I disconnected a Samsung Smart TV from the network (F series / cca. 2013). No particular reason other that there are SNB reports of network issues with Samsung TVs. It was connected to PLA (WiFi). It's smart features are outdated but the image quality is nice. I updated the Smart features with a Xiaomi 4k stick (Kodi, Youtube)

 

[jass]

To me it sounds like you have built an old school "franken-mesh" system, with a asus router and wifi in one part of the house, and a kind of tp-link pla wifi in the other part of the house, overlapping each other. And hope it will perform as a real mesh system...

I doubt you'll ever get satisfied with your current setup. In your case i would bite the apple and get another Aimesh unit instead of your tp-link wifi. Ideal same as the one you already have (ax88u). A lot is going on fw wise for the time beeing, and less chance of trouble if both your devices is on the same fw, at least the same branch.. Even though Asus claims you can mix and match as you like, but my gut says thats just asking for trouble.

If you decide to expand your wifi with an aimesh node, try If possible to disable the pla wifi, and use the pla to cable the new aimesh node.

Best of luck in your search for troublefree wifi and vacuming!

 

[ciprian.trofin]

Thank you for you kind words.
IIRC, the idea of using the same SSID (to allow me to roam free around house), came from TPLink PLA manual:

The TL-WPA4220 features a Wi-Fi Clone Button enabling Super Range Extension, which means it can automatically copy the SSID and Password of your router. In this way, TL-WPA4220 simplifies your Wi-Fi configuration and allows for seamless roaming within your home network.

My previous router was an RT-N66U.
I gave up on it because I needed more LAN ports. Also, the same problems I have today, were present during N66U era, but much less frequently.
Also, I you read my journal - errr... previous posts - you'll notice that about an year ago I had moments with 1-4 weeks of uptime. Not anymore.

My current attempt - I scheduled the following task, each day, at 5AM:

0 5 * * *  logger ARP stopped; ip link set arp off dev br0; sleep 2; ip link set arp on dev br0; logger ARP restarted #arp_restart#

 

[ciprian.trofin]

Crashed again. About 3 days uptime.

The log showed errors like:

rdpa_cpu_tx_port_enet_lan#213: rdpa_cpu_tx_port_enet_lan failed. rdd_rc=120 tx_rdd_error_count=23553

I found an older post from @ColinTaylor here: https://www.snbforums.com/threads/t...sb-drive-to-client-intermittently-fail.77879/ and I consider following the advice (even if only partially fits my use case as I don't use an USB stick on router).

I disconnected the PLA and will wait a few days (about a week) if router still crashes.
I everything is fine, I will reconnect the PLA, with WiFi disabled.

 

[ciprian.trofin]

I upgraded to Merlin 3004.388.8_4 and I used ethtool according @ColinTaylor's instructions.
Currently, the PLA is disconnected.
Limited testing shows better 2,4G coverage in 388.8_4 vs 384.19
1d14h uptime so far

 

[ciprian.trofin]

Well... funny stuff.

After 7 days of uptime, I attached the PLA unit, just the one directly connected to router. At day 10, I connected the other PLA unit (the WiFi access point). 3 days later (one hour ago), the Internet became unaccesible. I rebooted the router... and nothing.

The problem was the 12V power source for the ONT device. It broke, at the worst moment

I found another power source, I rebooted the router, entered Colin's code and... it's time to wait again.

 

[ciprian.trofin]

Crash again: 3 days after I connected the PLA (w/ AP enabled)

Mar 13 20:01:10 kernel: FPM Pool 0: invalid token 0x00009000 freed
Mar 13 20:01:10 kernel: FPM Pool 0: ISR timer is enabled. There could be multiple occurrences of the reported issue

The symptoms I noticed: the WiFi devices were not able to connect to network when the AP was PLA. Connections on 5GHz or 2.4GHz (directly to router) were OK.

As far as I know, there is no known fix for the issue printed in log (closed source program, or smth).

 

[ciprian.trofin]

8 days uptime (and still going).
Current "adjustments" are:
1. TCP segmentation offload is turned off, using @ColinTaylor's instructions; I'm using firmware 3004.388.8_4, ethtool is included, no need to install it from the Entware repository

ethtool -K br0 tx off sg off tso off gso off

2. STP is enabled via SSH; the powerline adapter is a Layer 2 device, and it might cause network loops

brctl stp br0 on

3. cronjob at 5AM to disable / enable ARP; the job is logged; I cannot really explain why I set this up; more or less is about the idea that restarting ARP might prevent some crashes related to my configuration

logger ARP stopped; ip link set arp off dev br0; sleep 2; ip link set arp on dev br0; logger ARP restarted

4. cronjob at 5:01AM to restart radio; the job is logged; I noticed that this might increase the interval between crashes (also, I could be wrong and just a coincidence)

logger Radio stopped; radio off; sleep 2; radio on; logger Radio restarted

5. cronjob at 5:02AM to simulate the removal and reconnection of a CAT5 cable from router (the cable from router to powerline adapter is connected to eth4); I don't remember where I got the idea from (SNB or somewhere else)

ethctl eth4 media-type 10HD; sleep 10; ethctl eth4 media-type 100FD

 

[ciprian.trofin]

It "semi-crashed" after 11 days (better than the regular 3). I solved the problem without restart, changing the router port for the powerline adapter.
Soft-reseting the port using ethctl gave no result.