CakeQOS CakeQoS-Merlin v2.0.0 alpha

[Mutzli]

Ideally I try to keep drops to a minimum - note minimum does not necessarily mean zero

If my drops are high do I have to specify a lower download/upload speed? Here are my stats:

Downloads:
Packets Sent    59,639,951
Sent    68.79 GB
Hash Indirect Hits    4,000,916
Hash Misses    291,082
Hash Collisions    0
Drops    19,342

Uploads:
Packets Sent    22,317,102
Sent    9.83 GB
Hash Indirect Hits    1,351,282
Hash Misses    251,869
Hash Collisions    0
Drops    1,823

 

[ttgapers]

If my drops are high do I have to specify a lower download/upload speed? Here are my stats:

Downloads:
Packets Sent    59,639,951
Sent    68.79 GB
Hash Indirect Hits    4,000,916
Hash Misses    291,082
Hash Collisions    0
Drops    19,342

Uploads:
Packets Sent    22,317,102
Sent    9.83 GB
Hash Indirect Hits    1,351,282
Hash Misses    251,869
Hash Collisions    0
Drops    1,823

Correct that is where I would start, lower those numbers and you may see better. Simple like Cake.

 

[slidermike]

ttgapers,
having installed the Alpha for pre-386.2 (386.1_2 in my case ac86u) should I be seeing a gui for cake-qos as shown in the examples?

I have installed and configured cake-qos via command line but thought I should see a gui interface.
My internet service is Docsis 450/25

Spoiler: current settings

I also came across a different buffer bloat testing page than dslreports.com

Bufferbloat and Internet Speed Test - Waveform

Is bufferbloat causing issues with your internet connection? Run this test to find out.

www.waveform.com

 

[Wistuplu]

If my drops are high do I have to specify a lower download/upload speed? Here are my stats:

Hello,
I see 1 drop every 2000 or 3000 packets. I wouldn't worry about it. Some packet drop is a feature of networking.

EDIT: Often, when packet drop is reported, it appears as a percentage, with e.g. one or 2 decimal figures most. So, see it as 0,1%, which appears probably less worrying.

Hoping to get the last round of Alpha Testing prior to pushing this to Develop.

2cts feedback.
In CLI and in the web ui, cake-qos presents itself now as just "2.0.0" even though it is currently alpha. And there has been more than one alpha available to us. My 2cts suggestion is to allow identifying the exact "version"/status. I can think of a suffix like "alpha 1", or later "beta 3" ;-), or "_RC1", or some sort of timestamp, or whatever. But it's odd to me that people will report some behaviour and it will be difficult to identify which codebase really produced that behaviour.
Anyway, thank you for cake-qos.
Best regards
W.

 

[dave14305]

having installed the Alpha for pre-386.2 (386.1_2 in my case ac86u) should I be seeing a gui for cake-qos as shown in the examples?

No, the WebUI is only going to support 386.2 or higher. The test is to ensure that the 2.0 script installs the correct legacy 1.0.x scripts for 386.1 users and 384.x users, so that we can safely push the 2.0.0 script to the master branch without completely stranding legacy users.

 

[New2This]

 

[Wade Coxon]

@New2This , it's a bit strange to be getting 0ms latency even for the unloaded test. Something not quite right with that screenshot. What detailed values were shown for those tests?

 

[abracadabra11]

@New2This , it's a bit strange to be getting 0ms latency even for the unloaded test. Something not quite right with that screenshot. What detailed values were shown for those tests?

I've seen similar results when I run the waveform bufferbloat from my Pixel 3. But the results are very different when run on my desktop (which is far more powerful and has a stronger WiFi signal and higher max bandwidth). Accordingly, I've ignored the Pixel results.

 

[Smokey613]

I wouldn't put too much trust into dslreports test. Some days I get a perfect A+ and other days I get a B or even C without changing any settings of cake.

Edit: Don't get me wrong, I still think it's generally a useful tool to check your settings in cake.

I got my first triple A+ on dslreports a few days ago, of course I have switched over to a 3 pack of eero pros with SQM turned on. BTW the eeros use Cake.

 

[slidermike]

So far so good. Still wish they would find a way to use multi-core. I hate not being able to get remotely near my 450Mbps download.

 

[TonyK132]

No, the WebUI is only going to support 386.2 or higher. The test is to ensure that the 2.0 script installs the correct legacy 1.0.x scripts for 386.1 users and 384.x users, so that we can safely push the 2.0.0 script to the master branch without completely stranding legacy users.

Should I try Cake 2.0 even though I'm on 384.19? Will I see a difference compared to 1.0.8?

 

[MvW]

Should I try Cake 2.0 even though I'm on 384.19? Will I see a difference compared to 1.0.8?

If I understood @dave14305 correctly, you can't. As long has you're on 384.19 you're limited to legacy version 1.0.8

From your own post:

The test is to ensure that the 2.0 script installs the correct legacy 1.0.x scripts for 386.1 users and 384.x users

 

[slidermike]

As a 386.1_2 user who used the "alpha" script to install cake, is seeing cake ver 1.08 correct?

I just want to make sure I should not be seeing the newer 2.0 cake though I am not on 386.2

Thank you

 

[MvW]

Either I don't understand @dave14305 his sentence or he should rephrase it, but reading

The test is to ensure that the 2.0 script installs the correct legacy 1.0.x scripts for 386.1 users and 384.x users

makes me assume that you need 386.2 to run CakeQos-Merlin 2.0.0.0 alpha. 386.1_2 is a minor release in the 386.1 series.

I could be completely wrong, but @dave14305 or @ttgapers will undoubtly drop by later to either confirm or correct me.

 

[dave14305]

As a 386.1_2 user who used the "alpha" script to install cake, is seeing cake ver 1.08 correct?

I just want to make sure I should not be seeing the newer 2.0 cake though I am no ton 386.2

Thank you

Yes, that is correct. The install should have told you your firmware wasn’t supported but installed the 386 branch version 1.0.8. Check the SCRIPT_BRANCH variable in the cake-qos file to confirm it now shows 386 for you instead of master.

 

[pdc]

Hello,
I see 1 drop every 2000 or 3000 packets. I wouldn't worry about it. Some packet drop is a feature of networking.

EDIT: Often, when packet drop is reported, it appears as a percentage, with e.g. one or 2 decimal figures most. So, see it as 0,1%, which appears probably less worrying.

So I'm not worrying, but I notice my download is around 1.7% dropped, with bandwidth set to 94% of speedtest (80 instead of 85). I'm going to try dropping back to 77.5 (91%) for a while and see how that goes... I hate "losing" the bandwidth, but responsiveness is likely more noticeable.

Upstream is much better, more like 0.08% (with over 20% of acks dropped!) though that is set to 92% (5.5 instead of 6).

 

[MvW]

So I'm not worrying, but I notice my download is around 1.7% dropped, with bandwidth set to 94% of speedtest (80 instead of 85). I'm going to try dropping back to 77.5 (91%) for a while and see how that goes... I hate "losing" the bandwidth, but responsiveness is likely more noticeable.

Upstream is much better, more like 0.08% (with over 20% of acks dropped!) though that is set to 92% (5.5 instead of 6).

Out of curiosity: based on which value do you calculate the percentage of drops? Maybe you can you could write up the formula. It would be nice to compare.

What would even be nicer is that @Jack Yaz hop on board and create some decent graphs because I'm more visually oriented and now chrony-nts is out an VoM(mit) 2.0 has been shelved indefinitely, I guess he needs a purpose for next weekend. So Jack, if you're reading this...

 

[Jack Yaz]

an VoM(mit) 2.0 has been shelved indefinitely, I guess he needs a purpose for next weekend. So Jack, if you're reading this...

it has?

 

[MvW]

it has?
View attachment 33111

Oooops.... I might have misread that In that case... maybe another weekend?

 

[TonyK132]

If I understood @dave14305 correctly, you can't. As long has you're on 384.19 you're limited to legacy version 1.0.8

From your own post:

If I understood @dave14305 correctly, you can't. As long has you're on 384.19 you're limited to legacy version 1.0.8

From your own post:

Right, but there is the process of verifying the version that might need to be checked by a 384.19 user. And Dave mentions 1.0.x, implying there might be something later than 1.0.8 either available or in the works.