FlexQoS FlexQoS - Why didn't it work?

[TheLyppardMan]

I'd had FlexQoS on my RT-AX88U for a while. Maybe I was expecting too much or maybe there was another reason, but it didn't stop my network virtually grinding to a halt yesterday while my son was doing a large upload to the Internet from his laptop, even though the speed settings for my Plusnet FTTC in Adaptive QoS were set below the recommended 95%. Most of the traffic seemed to be Net Control packets from what I remember (but don't take that as gospel, because my short-term memory is not brilliant these days). Has anyone else had a similar problem and if so, how did you resolve it? I'm using the latest release version of Merlin firmware on my router.

 

[Mutzli]

Exactly of problems like this I switched to Cake-QoS a while ago and never looked back. Merlin is working on brining Cake-QoS to the firmware as an option. It's available in the latest alpha release from yesterday.

 

[dave14305]

If it really was Net Control Packets for a file upload, then it is unlikely that FlexQoS was really active or fully applied. Or your upload bandwidth wasn't congested but your WiFi was. Or maybe you weren't using the fq_codel option in FlexQoS which gives better control and fairness than the Default configuration.

What are your speeds? You neglected to mention that important fact.

 

[thiggins]

@TheLyppardMan and everyone. Please don't use the "How To" prefix unless you are posting a How To procedure/method/process. This prefix is NOT to be used for asking how to do something.

Thanks,

 

[TheLyppardMan]

If it really was Net Control Packets for a file upload, then it is unlikely that FlexQoS was really active or fully applied. Or your upload bandwidth wasn't congested but your WiFi was. Or maybe you weren't using the fq_codel option in FlexQoS which gives better control and fairness than the Default configuration.

What are your speeds? You neglected to mention that important fact.

Hi Dave

I'm uploading speed tests for the last week for you. The big dip in the upload is when I noticed everything had slowed down. Even accessing the router GUI took much longer than usual and my daughter told me that she was experiencing slow internet as well.

 

[TheLyppardMan]

@TheLyppardMan and everyone. Please don't use the "How To" prefix unless you are posting a How To procedure/method/process. This prefix is NOT to be used for asking how to do something.

Thanks,

OK, sorry about that. I misunderstood the meaning.

 

[ttgapers]

Exactly of problems like this I switched to Cake-QoS a while ago and never looked back. Merlin is working on brining Cake-QoS to the firmware as an option. It's available in the latest alpha release from yesterday.

v1.0.8 updated last night as well. Next iteration will support @RMerlin baked cake in the new release(s)

 

[lilstone87]

v1.0.8 updated last night as well. Next iteration will support @RMerlin baked cake in the new release(s)

So will your updates be merged into RMerlin's builds? Because I'm guessing he will have to do it, and not something you can run a command to do any Cake updates.

 

[ttgapers]

So will your updates be merged into RMerlin's builds? Because I'm guessing he will have to do it, and not something you can run a command to do any Cake updates.

The script will allow for advanced settings that RMerlin does not expose natively through the GUI. He is baking the binaries for Cake into his releases which has held back the cakeqos scripts from a wider audience/install base. This should hopefully bring it on par with the other (more popular) offerings. Note cake is still the only QoS option that doesn't rely on any Privacy intrusions from having to accept the Trend Micro EULA.

Hope that helps.

 

[RMerlin]

Note cake is still the only QoS option that doesn't rely on any Privacy intrusions from having to accept the Trend Micro EULA.

Traditional QoS doesn't use the Trend Micro engine either (it's based on the original Tomato QoS code, and has evolved a bit in a different direction over time). Plus both with my firmware and the newer stock firmware you now get fq_codel instead of sfq.

 

[dave14305]

Plus both with my firmware and the newer stock firmware you now get fq_codel instead of sfq.

Stock now does fq_codel in Traditional? I think you mentioned somewhere that ASUS added fq_codel but it went over my head at the time. Did they do it their own way or use your code?

 

[RMerlin]

Stock now does fq_codel in Traditional? I think you mentioned somewhere that ASUS added fq_codel but it went over my head at the time. Did they do it their own way or use your code?

It was added at one point during the RC2 beta releases, following discussions I've had with them. Their code is slightly different because they didn't want to support multiple schedulers (which my code still supported at the time), or the overhead parameters. They basically just replaced sfq with fq_codel in the existing code and did some tweaks to quantums, so there isn't much to it.