FlexQoS RT-AX86U (Asus Merlin 388.2_2) || FlexQoS Not Starting

[kazu97]

Hello All,

I am having a bit of trouble with FlexQoS. I keep getting TC Modification Delayed Start, into it giving the same error every three minutes. Normally I would ignore this, but every time this this error pops up, I notice a spike in latency when I am playing a game. It is very annoying and I am not sure why it is happening. Attached are the logs. I tried updating it and hit the debugger twice.
Debug -> update -> restart -> debug

I had to attach the debug file instead of posting as a spoiler as I ran out of character limit.


Any help would be appreciated. Thanks!

 

[ColinTaylor]

I suggest you change your title to say "FlexQoS" rather than "Adaptive QoS" as that is what you are asking about.

Post the complete system log as well so we can see the error message you're referring to in context with the other messages.

 

[dave14305]

You should also remove devices from the Gear Accelerator feature. It conflicts with FlexQoS. Or disable FlexQoS.

Please post the output of:

tc qdisc show root
tc filter show dev br0
grep FlexQoS /tmp/syslog.log-1 /tmp/syslog.log

 

[kazu97]

You should also remove devices from the Gear Accelerator feature. It conflicts with FlexQoS. Or disable FlexQoS.

Please post the output of:

tc qdisc show root
tc filter show dev br0
grep FlexQoS /tmp/syslog.log-1 /tmp/syslog.log

I think this was it, I had a device in gear accelerator that I forgot about (fingers crossed it wasn't just a coincidence and I have bad timing).

Thank you!

If it comes around again I will post the logs with those commands

 

[dave14305]

If it comes around again I will post the logs with those commands

I don’t think Gear Accelerator was the cause, just something I noticed in your debug output.

 

[kazu97]

I don’t think Gear Accelerator was the cause, just something I noticed in your debug output.

youre right. I checked again this morning and noticed it popped up again. Ran the code and got the output, attached.

Cheers,

kazu

 

[dave14305]

youre right. I checked again this morning and noticed it popped up again. Ran the code and got the output, attached.

Cheers,

kazu

Helpful. I see the use of bond1 exposes a long-standing flaw in FlexQoS upload interface detection.

I’ve pushed a fix that in theory should work. Please update to v1.3.3 and test it out.

Thanks.

 

[kazu97]

Awesome, glad I could be of use.

Will try it out right now, will respond later after a night of gaming.

 

[kazu97]

Just tested it, seems to be working well. Thanks for your work! To note, I was doing some bufferbloat tests via waveform last night and the min/max bandwidth was not being respected when doing them (which is to be expected since FlexQoS was not starting). Just tested it now and got an A+, with the bandwidth limits being respected. +25 download/+9 upload average -> +4 download/+1 upload average, so pretty much perfect. Will test again later.