QoSmate: (Yet Another) Quality of Service Tool for OpenWrt

[lytr]

@Hudra

Are you aiming to create a database of QoS performance metrics per router, or what exactly should be measured?

My idea is a script that would be able to determine the performance of the device in some way.

@Hudra

For example, are you focusing on throughput alone, or throughput plus latency, CPU utilization, or all of them combined?

It could be just a throughput test to start with, and some other metrics could be added later.

@Hudra

If you create a test environment using netns and iperf3 as in your example script to measure QoS performance the test will likely be more synthetic and may not accurately reflect the real-world conditions when applying QoS to the WAN-facing interface.

There is no problem with this, it is more about estimating the capabilities of the device than the actual results.

@Hudra

If I remember correctly, betterspeedtest does something similar and could potentially be integrated into qosmate. However, this is not something you couldn't already test with the waveform bufferbloat test or the cloudflare speedtest.

I am thinking about a test that would be performed internally without the need for communication with external services.

[Hudra]

@lytr

the performance of the device in some way.

Hmm yeah, but that’s exactly the question: What kind of performance are you trying to measure and what problem are you actually trying to solve with it?

If we implement a synthetic benchmark what kind of conclusions can we realistically draw from it?

For example, if your router proves capable of traffic shaping at 1 Gbps in a test, but fails to do so in a real-world scenario then what’s the point of the test?

[lytr]

@Hudra

Hmm yeah, but that’s exactly the question: What kind of performance are you trying to measure and what problem are you actually trying to solve with it?

If we implement a synthetic benchmark what kind of conclusions can we realistically draw from it?

I mean estimating the maximum performance of the processor. By running the same test on different devices, you can create a database of results.

Such a database can help you choose a device. Currently, you can find very diverse information from different users.

@Hudra

For example, if your router proves capable of traffic shaping at 1 Gbps in a test, but fails to do so in a real-world scenario then what’s the point of the test?

Real results may differ significantly, but this depends on many factors.

[Knomax]

Hi..i want to thanks again @Hudra for his work and @antonk for update system.
Until now even in my low speeds i get in Waveform bufferbloat test 1-3ms for download and stable 0 ms in upload direction!!
Also my rule for game console is only for UDP of course and was seeing until now that if i wasnt playing "live" and seeing something in games menu etc trophy's,medals,tanks (i play wot) and such things game menu wasnt "fast".
Now i think something has to do with "Enable Initial TCP Down-Prioritization" and "Enable Sustained TCP Down-Prioritization" game menu is much more faster ..changing tabs and so on...maybe it uses TCP traffic?Or maybe it is "placebo" from me i dont know.
For sure i "see" an improvement .
Again thank you guys.

[Hudra]

@Knomax

Now i think something has to do with "Enable Initial TCP Down-Prioritization" and "Enable Sustained TCP Down-Prioritization" game menu is much more faster ..changing tabs and so on...maybe it uses TCP traffic?Or maybe it is "placebo" from me i dont know.
For sure i "see" an improvement .

What exactly do you mean by that? Are you seeing an improvement when these two settings are enabled or disabled?

[Knomax]

@Hudra

What exactly do you mean by that? Are you seeing an improvement when these two settings are enabled or disabled?

Both are enabled and see improvement...as i said maybe it is "placebo" from me?

[Hudra]

Then it’s probably more of a placebo effect, since those two settings have always been active before. The only thing that changed is that they can now be disabled.

[Knomax]

In statistics tab i have some "Drops" and the same number in "Overlimits" for "High Priority (Realtime)" class in normal class has only "drops" but i dont care for this class.
I change the speeds in settings but didnt change anything...i use "red" for Queue Discipline.
Can i do something about it?
And i understand "drop"....overlimit what means?

[moeller0]

@Knomax

And i understand "drop"....overlimit what means?

Overlimits with qdiscs typically just means that the qdisc actually engaged, this is not an error it is really just an information if you ask, did my realtime qdisc actually ever do something?