CakeQOS CakeQoS-Merlin v2.0.0 alpha

[askan7]

Can anyone help me with DSCP tagging? I would like to tag all UDP traffic soo it goes to the Voice tin. And maybe try per port tagging if that isn't enough.

 

[heysoundude]

I'm actually having an issue trying to input my up and down speeds: the interface locks up when I click on them. EDIT - needs doing in the QoS tab
AC86, just updated to 386.2 45 mins ago from 386.1_2. needed to reboot a couple of times before I got it to connect to my ISP, then I went in and removed the old cake and copypasta'd the instruction in post #1 here (the re-re was FAST).
and now it seems I'm missing my unbound and scMerlin tabs...
Hmmm...standby, I'm going digging. will update as I fix things
UPDATE - after forcing updates of scMerlin and rebooting a couple (3?) times, all is right again in the GUI

 

[askan7]

Ubuntu Manpage: CAKE - Common Applications Kept Enhanced (CAKE)

manpages.ubuntu.com

From that page:

wash

Traffic entering your diffserv domain is frequently mis-marked in transit from the
perspective of your network, and traffic exiting yours may be mis-marked from the
perspective of the transiting provider.

Apply the wash option to clear all extra diffserv (but not ECN bits), after priority
queuing has taken place.

If you are shaping inbound, and cannot trust the diffserv markings (as is the case for
Comcast Cable, among others), it is best to use a single queue "besteffort" mode with
wash.

What's the point os using wash with besteffort? Since you only have a single tin DSCP markings should matter right? I thought wash was only useful when using diffserv soo packets don't end up on the wrong tin.

 

[dave14305]

What's the point os using wash with besteffort? Since you only have a single tin DSCP markings should matter right? I thought wash was only useful when using diffserv soo packets don't end up on the wrong tin.

wash only clears DSCP after the traffic has been tinned. Strange but true.

 

[heysoundude]

What's the point os using wash with besteffort? Since you only have a single tin DSCP markings should matter right? I thought wash was only useful when using diffserv soo packets don't end up on the wrong tin.

I included the link so that you might RTFManpage - it helps. for everything. try it. manpages are a really good resource and they're often not as technical as you might fear

 

[askan7]

wash only clears DSCP after the traffic has been tinned. Strange but true.

Soo, what's the point of wash? I don't see an use for it.
I'm not using wash and when I open wireshark all the packets are markets CS0, with some CS6 (icmp packets) I don't think my ISP is marking packets at all.
The CS6 packets I see are sent from router itself, and are always market CS6 even with wash. EVERYTHING else is market CS0 wash or no wash.

I would like to know how could I mark packets with DSCP, for example to apply priority to a game by port number.

 

[Wade Coxon]

Soo, what's the point of wash? I don't see an use for it.

It may still be useful to wash the marks so that any additional switches and routers behind your router do not act on the dscp marks. Cake is designed to be used on all sorts of enterprise class equipment too, not just home routers.

For us, I don't see a whole lot of point in it.

 

[dave14305]

WiFi WMM also uses DSCP marks so it might help in certain situations (Comcast seems to be the poster boy). It also means you could manipulate DSCP internally to get the tin you want, but wash it out before sending it onward to the internet or to the LAN.

 

[randomName]

Thanks @Wade Coxon for getting me hip to this. I guess I'll ask all my questions here.

For a dirty update, first uninstall cake 1.0.8, install 386.2_0, run the addon script via putty, reboot router. Do I need entware anymore? I basically just use Skynet and Cake

 

[cplay]

@dave14305 - Is there something I'm missing with cake?

Whatever bandwidth limits I set in manual options speedtests and waveform tests get nowhere near that value AND bufferfloat is substantially higher than compared to when I run flexqos (it also performs substantially worse when gaming and when streaming from multiple devices)?

I haven't installed this cake 2 app yet, but is there something I'm missing when it comes to both the above issues on my 100,50 4g line.

I set cake to 59/31 exactly the same as I set flex but flex performs substantially better.

So, I guess the question is: What settings should I select with your cakeqos addon to emulate the closest setup to flexqos?

Hope you are well - have a lovely weekend mate!

 

[askan7]

@dave14305 - Is there something I'm missing with cake?

Whatever bandwidth limits I set in manual options speedtests and waveform tests get nowhere near that value AND bufferfloat is substantially higher than compared to when I run flexqos (it also performs substantially worse when gaming and when streaming from multiple devices)?

I haven't installed this cake 2 app yet, but is there something I'm missing when it comes to both the above issues on my 100,50 4g line.

I set cake to 59/31 exactly the same as I set flex but flex performs substantially better.

So, I guess the question is: What settings should I select with your cakeqos addon to emulate the closest setup to flexqos?

Hope you are well - have a lovely weekend mate!

I have yet to try with 4G, but cake implementation on 386.2 gave me better results than cake 1.0.8. You should try it. I have to try with 4G, I have a 12gb data per month to test. It gives me around 50/6 but much higher ping than adsl.

Also with cake you have to set limits a bit higher than flexqos/adaptive qos to get the same bandwidth.

 

[cplay]

I have yet to try with 4G, but cake implementation on 386.2 gave me better results than cake 1.0.8. You should try it. I have to try with 4G, I have a 12gb data per month to test. It gives me around 50/6 but much higher ping than adsl.

Also with cake you have to set limits a bit higher than flexqos/adaptive qos to get the same bandwidth.

Yeah, I tried it on the latest build no difference.

my 4g is solid 39ms 1/2ms jitter and speeds barely change (I live rural so) - I just can’t get cake anywhere near as functional for my line compared to flexqos

 

[askan7]

Yeah, I tried it on the latest build no difference.

my 4g is solid 39ms 1/2ms jitter and speeds barely change (I live rural so) - I just can’t get cake anywhere near as functional for my line compared to flexqos

I just tested cake with 4G and I get really unstable download speeds if I set a limit. I guess cake drops a lot of packets to keep latency low. Automatic = stable speeds but 1000+ms latency.
Maybe I could improve this by setting proper overhead (I used 0). Anyway 4G is not in any way stable. I still prefer ADSL even with much lower bandwidth. I'm getting around 30-40/6 with 4G. My Phone gets 50/4 on the same place and ISP but latency and jitter is even higher.

meter.net/ping-test gives me +15ms on average compared to ADSL and 5-6ms jitter compared to 0.5-1ms.

 

[zackattack784]

I'm currently running (and have been for months) the cake-qos script. I'm very happy with the performance. Any reason to switch to merlin's built in solution?

 

[Wade Coxon]

I'm currently running (and have been for months) the cake-qos script. I'm very happy with the performance. Any reason to switch to merlin's built in solution?

The built in solution works just the same and is less "hacky" (ie. it has a web configuration screen and messing around at the command line is not required), and does not require extra entware packages. I don't see any reason for not using the built-in solution.

 

[MvW]

The built in solution works just the same and is less "hacky" (ie. it has a web configuration screen and messing around at the command line is not required), and does not require extra entware packages. I don't see any reason for not using the built-in solution.

As you're posting your question in the CakeQos-Merlin Alpha v2.0.0.0 thread, I would like to add to Wades' post that the UI of the built-in solution is less configurable then the alpha in the first post. True, messing around with the command line isn't required anymore, but you'll still be editing a config file with Merlins' solution, while the alpha offers a UI with far more configurable options than the built-in solution. RMerlin explained that this is on purpose, he just wants a simplified version which basically works for everyone, so the possibilities to tweak it to your own personal preferences are less easy accessible.

 

[zackattack784]

Thanks, I'll give Merlin's implementation a go with the 2.0 Alpha and see if I notice any differences.

 

[Morris]

Just noticed when using Cake, if I download from STEAM while using Teamspeak, the entire teamspeak lags. Since I host the teamspeak 3 server on a spare pc on the same network, apparently it saturates the network so much that everyone lags and they all say they hear "robotic" voices from the lag. So it seems CAKE doesn't resolve these types of issues?

When a real time application saturates the network, QOS can't do anything to fit more in the pipe. The only fix is to provision more bandwidth.

 

[Morris]

Can anyone help me with DSCP tagging? I would like to tag all UDP traffic soo it goes to the Voice tin. And maybe try per port tagging if that isn't enough.

You should be running flex QOS, this is not the objective of Cake

 

[Morris]

What's the point os using wash with besteffort? Since you only have a single tin DSCP markings should matter right? I thought wash was only useful when using diffserv soo packets don't end up on the wrong tin.

To prove the sane decision or Eric Merlin not to expose these features!