Bufferbloat only on 2.4GHz

[lzkill]

Hi folks. I have two RT-AC68R in a mesh setup (386.1_2). My Internet connection is a symmetrical 100 Mbps fiber. On the main router I've enabled QoS and FlexQoS 1.2.2 (amtm). The configuration is adaptative, bandwidth manually set to 90 Mbps, fq_codel selected. The traffic classes are ordered as suggested by the FlesQoS docs (File downloads with the lowest priority).

I can confirm the ping to 8.8.8.8 keeps low on the main router even when someone is downloading a file or making a speed test. What I don't understand is why dslreports give me a bad bufferbloat grade when I run its test from a 2.4 Ghz device. If a run the site's test using the 5GHz band the result is always A+. What am I missing here? The speeds I get on 5GHz is 90 Mbps and 60 Mbps on 2.4 GHz.

I've done all the tests connected to the main router, so I can ignore any latency added by the AiMesh node. I must also add that my ISP does chat, so my main router is not fully exposed to the Internet.

Thanks guys!

 

[ttgapers]

Might simply be congestion/interference on your 2.4 band. In general, I do my bufferbloat tests from a wired device for a more consistent result which also eliminates any wireless interference.

 

[bbunge]

Hi folks. I have two RT-AC68R in a mesh setup (386.1_2). My Internet connection is a symmetrical 100 Mbps fiber. On the main router I've enabled QoS and FlexQoS 1.2.2 (amtm). The configuration is adaptative, bandwidth manually set to 90 Mbps, fq_codel selected. The traffic classes are ordered as suggested by the FlesQoS docs (File downloads with the lowest priority).

I can confirm the ping to 8.8.8.8 keeps low on the main router even when someone is downloading a file or making a speed test. What I don't understand is why dslreports give me a bad bufferbloat grade when I run its test from a 2.4 Ghz device. If a run the site's test using the 5GHz band the result is always A+. What am I missing here? The speeds I get on 5GHz is 90 Mbps and 60 Mbps on 2.4 GHz.

I've done all the tests connected to the main router, so I can ignore any latency added by the AiMesh node. I must also add that my ISP does chat, so my main router is not fully exposed to the Internet.

Thanks guys!

Likely that the 2.4 GHZ throughput is less than the 100/100 FIOS. If the 2.4 connection is going through the mesh node it could be worse especially if the node is on WIFI backhaul. Browser add blockers mess up the speed test, too.
I have a 100/100 FIOS, set the QOS manual at 100/100 and do not use FlexQOS (right now) and have A+, A+, A DSL reports
If the test via wired client is good, don't worry about the rest.