New RT-AX86U - Strange Behavior - Questions

[Dan-in-VA]

I don't believe it's good practice to reload a setting's file after a factory reset, if that settings file didn't come from that same firmware version. From what I've read, the router needs to be manually configured after a f/w up/down/merlin-grade + full reset (unfortunately).

Also, if you want a stable non-merlin f/w, then try the stock 3.0.0.4.386_46061. I've been running it for 5 months with no issues, and many Apple devices, new and old.

I took the setting file backup before I upgraded to the 3/25 version. So I restored the setting file taken in the same version.

If I had several hours with nothing to do, I could write out a procedure with all my settings. I used to do that, but stopped about 14 years ago.

Bottom line: I experienced the exact issue the OP did and this was an immediate 100% resolution. And—I would have stayed on Merlin if it didn’t have the SMB issue (reported elsewhere on the forum).

 

[SAL9K]

I took the setting file backup before I upgraded to the 3/25 version. So I restored the setting file taken in the same version.

If I had several hours with nothing to do, I could write out a procedure with all my settings. I used to do that, but stopped about 14 years ago.

Bottom line: I experienced the exact issue the OP did and this was an immediate 100% resolution. And—I would have stayed on Merlin if it didn’t have the SMB issue (reported elsewhere on the forum).

The OP has an AX86U, not an AC86U.

 

[seadragon]

RSSI is the received signal strength indicator. So the strength of the signal received by the router from the client will be different than that of the client from the router. As the transmission power of client devices is usually much lower than that of the router the RSSI figure shown by the router is typically lower than that shown by the client.

Hi Colin... just a follow up question about RSSI if I may. Now that I understand the values will be different from the router and client point of view, does one use the lower value of the client device to evaluate if a mesh node is needed? For example, a device that reads RSSI of -65 signal from the router and the router reads -85 signal from the client device would indicate a node may be appropriate?

 

[ColinTaylor]

I would say that generally speaking that should be the case. In your example -85 is the lower (weaker) signal and -65 is the stronger. Whichever way around the relative signal strengths are, the link is going to be problematic if there's a very weak signal in one direction. All data transfers require bidirectional communication, even if the volume of data being transferred in each direction is asymmetric (which it usually is).

With all that said, this is really something you just have to experiment with yourself and see what works in your environment for you use cases. You could have a weak but stable WiFi connection, the only downside being lower throughput. That might be perfectly acceptable for a mobile phone that's just checking emails, but not acceptable for a laptop computer.

 

[seadragon]

Just ran another 3 speed tests this morning back to back. First run was perfect the the second and third were slower. I again verified the router is getting full speed so it’s definite a wireless thing.

 

[seadragon]

So, I've tried several devices now and they all cannot reach 300 Mbps speed using wi-fi. If I connect the same devices to the router via ethernet I get full speed every time. I'm beginning to suspect the router is defective. It's not like I have Gig internet. My service is 300 down 20 up. I should be easily hitting this with wi-fi. I'm going to take some of my devices to a friend's place later today who has Gig fibre service and a TPLink Archer router. If I get good speeds there, I'll know my wi-fi is at fault.

 

[ColinTaylor]

Just ran another 3 speed tests this morning back to back. First run was perfect the the second and third were slower. I again verified the router is getting full speed so it’s definite a wireless thing.

How far away from the router were you when you ran the tests? For example, the -85dBm signal strength you mentioned in post #43 is extremely weak, borderline unusable.

 

[seadragon]

How far away from the router were you when you ran the tests? For example, the -85dBm signal strength you mentioned in post #43 is extremely weak, borderline unusable.

I am about 8 feet in front of the router when testing. The -85 reading was out in my attached garage.

 

[SAL9K]

Do and iperf3 test from a macbook to the AX86U. Doing an experiment at another site tells you nothing.

 

[seadragon]

Do and iperf3 test from a macbook to the AX86U. Doing an experiment at another site tells you nothing.

Well, it told me that nothing is wrong with my devices. So I know it's now something with the router. I just go back from there and all my devices were hitting 900 Mbps no problem. I'll try the iPerf next. I have an M1 MacBook Pro and noticed iPerf is for Intel but I'll try running it.

 

[SAL9K]

Well, it told me that nothing is wrong with my devices. So I know it's now something with the router. I just go back from there and all my devices were hitting 900 Mbps no problem. I'll try the iPerf next. I have an M1 MacBook Pro and noticed iPerf is for Intel but I'll try running it.

All Apple devices hit 900Mbps no problem? What iPhone do you have that can hit ~900Mbps? From my experience it's difficult to get to the 1Gbps w/out 160MHz channels.

 

[seadragon]

All Apple devices hit 900Mbps no problem? What iPhone do you have that can hit ~900Mbps? From my experience it's difficult to get to the 1Gbps w/out 160MHz channels.

I was using my iPad Air 4 and M1 Max 16" MacBook Pro. Sorry, the 900 was when connected via Ethernet (both iPad and MacBook Pro). I was getting around 650 - 750 using wi-fi on both devices. Still a lot more than what I'm getting from my router.

I just did an iPerf test with my MacBook Pro connected to the router via ethernet and running as server. I then used my wife's M1 MacBook Air to run the client. I hope I did this test right as I've never used iPerf before. Here are the results:

 

[SAL9K]

I was using my iPad Air 4 and M1 Max 16" MacBook Pro. Sorry, the 900 was when connected via Ethernet (both iPad and MacBook Pro). I was getting around 650 - 750 using wi-fi on both devices. Still a lot more than what I'm getting from my router.

I just did an iPerf test with my MacBook Pro connected to the router via ethernet and running as server. I then used my wife's M1 MacBook Air to run the client. I hope I did this test right as I've never used iPerf before. Here are the results:

That does look slow, even for an "upload" speed test, hmm (unless that's an older Macbook Air). Try a "download" test, E.g. same test in the reverse direction by adding '-R' to the end of the iperf3 command run on the Macbook Air. In general, it might be best to run your slowest/oldest device (E.g. Macbook Air) as the server connected via 1GbE, and your fastest/newest AX devices as the WiFi clients (Macbook Pro, iPad Air 4), etc. BTW, I see a few iperf3 apps in the iOS app store, for $1.99 USD. If you eventually try running iperf3 directly on the AX86U (as server), then you can run iperf3 tests all day long and not eat into your ISP data cap!

 

[seadragon]

That does look slow, even for an "upload" speed test, hmm (unless that's an older Macbook Air). Try a "download" test, E.g. same test in the reverse direction by adding '-R' to the end of the iperf3 command run on the Macbook Air. In general, it might be best to run your slowest/oldest device (E.g. Macbook Air) as the server connected via 1GbE, and your fastest/newest AX devices as the WiFi clients (Macbook Pro, iPad Air 4), etc. BTW, I see a few iperf3 apps in the iOS app store, for $1.99 USD. If you eventually try running iperf3 directly on the AX86U (as server), then you can run iperf3 tests all day long and not eat into your ISP data cap!

Thanks again for your help here. I just ran another test with the -R switch and these numbers look better. However, there was the occasional slower performance (such as the second test below). Now I don't know what to think. LOL.

 

[SAL9K]

Well, some variability over WiFi is to be expected (although yours are going btwn 200-800, which is very wide), unless you have a very dedicated RF testbed for such things (like @thiggins might have). Those are excellent WiFi speeds, at the upper end, perhaps try a few more runs in different positions to convince yourself things are ok. The upload deficit might be do to a power limitation on the TX device, I.e. the Apple device. But, those RX/TX speeds should easily cover your 300/30 ISP plan, so I'm not really sure why you get such low/inconsistent Ookla speedtest.net results. Maybe try https://fast.com/?

 

[seadragon]

Well, some variability over WiFi is to be expected (although yours are going btwn 200-800, which is very wide), unless you have a very dedicated RF testbed for such things (like @thiggins) might have. Those are excellent WiFi speeds, at the upper end, perhaps try a few more runs in different positions to convince yourself things are ok. The upload deficit might be do to a power limitation on the TX device, I.e. the Apple device. But, those RX/TX speeds should easily cover your 300/30 ISP plan, so I'm not really sure why you get such low/inconsistent Ookla speedtest.net results. Maybe try https://fast.com/?

Thanks. I did also try fast.com but I get the same results as Ookla. It's all very strange.

 

[SAL9K]

fast.com has a "Show more info" button, that further reports latency and upload speeds. Please try that, and post results here. I'm going to assume that you're running all of these Ookla/fast.com/iperf3 tests from the same physical (nearby) location.

 

[seadragon]

Thanks. I just ran fast.com with these results. I ran it a few times actually and the highest download was 220 and the lowest was 180.

And all of my testing is done from a table at this location. This is the distance to the router.

 

[SAL9K]

Here's what I get from a 2019 Macbook Pro (WiFi AC only, 80MHz ch), about 15ft from router, line-of-sight. Pretty symmetrical up/down, but seems to vary btwn 500-700 Mbps from run to run.

$ iperf3 -c 10.1.1.10
Connecting to host 10.1.1.10, port 5201
[  5] local 10.1.1.148 port 61572 connected to 10.1.1.10 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  57.7 MBytes   484 Mbits/sec             
[  5]   1.00-2.00   sec  72.3 MBytes   607 Mbits/sec             
[  5]   2.00-3.00   sec  87.3 MBytes   732 Mbits/sec             
[  5]   3.00-4.00   sec  86.5 MBytes   726 Mbits/sec             
[  5]   4.00-5.00   sec  78.4 MBytes   658 Mbits/sec             
[  5]   5.00-6.00   sec  87.1 MBytes   730 Mbits/sec             
[  5]   6.00-7.00   sec  89.1 MBytes   746 Mbits/sec             
[  5]   7.00-8.00   sec  83.7 MBytes   703 Mbits/sec             
[  5]   8.00-9.00   sec  84.1 MBytes   706 Mbits/sec             
[  5]   9.00-10.00  sec  83.9 MBytes   703 Mbits/sec             
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec   810 MBytes   680 Mbits/sec                  sender
[  5]   0.00-10.01  sec   808 MBytes   677 Mbits/sec                  receiver

iperf Done.
$ iperf3 -c 10.1.1.10 -R
Connecting to host 10.1.1.10, port 5201
Reverse mode, remote host 10.1.1.10 is sending
[  5] local 10.1.1.148 port 61574 connected to 10.1.1.10 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  49.0 MBytes   411 Mbits/sec             
[  5]   1.00-2.00   sec  68.3 MBytes   573 Mbits/sec             
[  5]   2.00-3.00   sec  68.6 MBytes   575 Mbits/sec             
[  5]   3.00-4.00   sec  85.3 MBytes   715 Mbits/sec             
[  5]   4.00-5.00   sec  79.8 MBytes   669 Mbits/sec             
[  5]   5.00-6.00   sec  86.2 MBytes   723 Mbits/sec             
[  5]   6.00-7.00   sec  94.0 MBytes   788 Mbits/sec             
[  5]   7.00-8.00   sec  91.2 MBytes   765 Mbits/sec             
[  5]   8.00-9.00   sec  89.8 MBytes   754 Mbits/sec             
[  5]   9.00-10.00  sec  84.4 MBytes   708 Mbits/sec             
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.02  sec   799 MBytes   669 Mbits/sec    0             sender
[  5]   0.00-10.00  sec   797 MBytes   668 Mbits/sec                  receiver

iperf Done.

My fast.com/Ookla tests are all over the place too, and definitely not as consistent as iperf3. I only have a 300/10 ISP plan

 

[SAL9K]

Did you say that you checked to make sure QoS was disabled, "Adaptive QoS"->QoS? I'm not sure if QoS affects speed tests, but I've always turned QoS off.