CakeQOS CakeQOS-Merlin

[ttgapers]

Don't worry about what the ISP marketing claims to have sold you, just worry about what your tests reveal .
Run a speed test with no QoS enabled, and pick the values that seem to beset represent your maximum steady state transfer rates for upload and download. Use those values as the maximum, and then apply your percentages on those tested values.

For example, when I run the DSLReports speed test on my link, I get around 43Mbps for download and a nasty C rated bufferbloat result with peaks in the hundreds of ms. I applied the recommended 85% QoS overhead and come to around 29Mbps and enable cake. Vastly improved, but bufferbloat is at A level with average bloat 20ms above idle. Dropping down to 28500kpbs results in A+ bufferbloat and average bloat within 5ms of idle. Perfect .

And thank goodness we don't need to wait 5 minutes between QoS adjustments like with the TrendMicro scripts!!

Awesome, try taking it up to 95% and see if you still get As
Note, I am just curious...

 

[Amwjujo]

I tested CAKE since yesterday on my 350/50 fiber connection at 95% but in my spdMerlin report I started to have this pikes on the speed download test - didn't had the before - but the reports from dslreports shows my bufferbloat on A and quality on A+.
Are this fluctuations normal and i am getting those readings because the way CAKE is prioritising the download?

 

[SomeWhereOverTheRainBow]

I tested CAKE since yesterday on my 350/50 fiber connection at 95% but in my spdMerlin report I started to have this pikes on the speed download test - didn't had the before - but the reports from dslreports shows my bufferbloat on A and quality on A+.
Are this fluctuations normal and i am getting those readings because the way CAKE is prioritising the download?

looks normal because Cake is adjusting for performance at whatever you set it at. I recommend testing it at 195/50 to see if you see the same spikes. The Cake is only recommended to support a max speed of ~200 to 230. I recommend testing it at 195. I know this is not the max of what you are paying for with your ISP, but you will notice alot less CPU usage and better client overall stability and performance.

 

[SomeWhereOverTheRainBow]

I tested CAKE since yesterday on my 350/50 fiber connection at 95% but in my spdMerlin report I started to have this pikes on the speed download test - didn't had the before - but the reports from dslreports shows my bufferbloat on A and quality on A+.
Are this fluctuations normal and i am getting those readings because the way CAKE is prioritising the download?

If you get decent performance, at 195/50- try updating to 215/50

 

[maghuro]

v0.0.4 released:

- added status option which will show you if cake is running and what your download/upload parameters are
- update now checks and downloads latest script, so in theory from this point on you just need to run /jffs/scripts/cake-qos update {ac86u|ax88u} to get the latest binaries/script
- added check to start cake only when Entware is started for consistent start-ups (thanks @JGrana)
- added checks to not double-start cake (e.g. if you start cake when it's running, it will stop cake and start cake to avoid the "RTNETLINK answers: File exists" message)

https://github.com/ttgapers/cakeqos-merlin#cakeqos-merlin

We weren't able to implement @Martineau's code contribution on model autodetection (but many thanks!); we will wait for someone to hopefully compile cake for more models

IMHO instead of services-start, it should go to nat-start.
Otherwise, I think cake won't survive a firewall restart.
And nat-start is called also after a firewall restart.

@Xentrk can you share your expertise regarding this? You had the same problem with vpnrouting from your x3mRouting.

 

[SomeWhereOverTheRainBow]

IMHO instead of services-start, it should go to nat-start.
Otherwise, I think cake won't survive a firewall restart.
And nat-start is called also after a firewall restart.

@Xentrk can you share your expertise regarding this? You had the same problem with vpnrouting from your x3mRouting.

So it is not question of how, but when. I would place my bets on firewall-start.

 

[Amwjujo]

If you get decent performance, at 195/50- try updating to 215/50

Those readings are from 280/48 setup.
Maybe I can decrease that a but and see what happens.

 

[brummygit]

IMHO instead of services-start, it should go to nat-start.
Otherwise, I think cake won't survive a firewall restart.
And nat-start is called also after a firewall restart.

@Xentrk can you share your expertise regarding this? You had the same problem with vpnrouting from your x3mRouting.

Just rebooted on version v0.0.4 and it doesn't seem to survive the reboot.
After reboot

After running an enable

Ignore the speeds as I am currently on a slow link with other traffic competing with me

EDIT:

Looking at the router log file it seems to think it was running as there is a stop before my enable

Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:27:16 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbi | besteffort
Jun 22 09:27:16 kernel: Cpuidle Host Clock divider is disabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:27:37 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbi | besteffort
Jun 22 09:27:37 kernel: Cpuidle Host Clock divider is disabled
Jun 22 09:29:45 WLCEVENTD: Deauth_ind 92:49:7F:D7:D1:7E
Jun 22 09:29:45 WLCEVENTD: Auth 92:49:7F:D7:D1:7E
Jun 22 09:29:46 WLCEVENTD: ReAssoc 92:49:7F:D7:D1:7E
Jun 22 09:29:46 WLCEVENTD: Disassoc 92:49:7F:D7:D1:7E
Jun 22 09:29:49 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:29:50 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:29:50 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:29:50 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbit | besteffort bridged-ptm ack-filter
Jun 22 09:29:50 kernel: Cpuidle Host Clock divider is disabled

 

[TechTinkerer]

Just rebooted on version v0.0.4 and it doesn't seem to survive the reboot.
After reboot

After running an enable

Ignore the speeds as I am currently on a slow link with other traffic competing with me

EDIT:

Looking at the router log file it seems to think it was running as there is a stop before my enable

Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:27:16 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:27:16 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbi | besteffort
Jun 22 09:27:16 kernel: Cpuidle Host Clock divider is disabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:27:37 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:27:37 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbi | besteffort
Jun 22 09:27:37 kernel: Cpuidle Host Clock divider is disabled
Jun 22 09:29:45 WLCEVENTD: Deauth_ind 92:49:7F:D7:D1:7E
Jun 22 09:29:45 WLCEVENTD: Auth 92:49:7F:D7:D1:7E
Jun 22 09:29:46 WLCEVENTD: ReAssoc 92:49:7F:D7:D1:7E
Jun 22 09:29:46 WLCEVENTD: Disassoc 92:49:7F:D7:D1:7E
Jun 22 09:29:49 hodgetts: CakeQOS-Merlin v0.0.4 Stopping
Jun 22 09:29:50 kernel: Cpuidle Host Clock divider is enabled
Jun 22 09:29:50 hodgetts: CakeQOS-Merlin v0.0.4 Enabled
Jun 22 09:29:50 hodgetts: CakeQOS-Merlin v0.0.4 Starting - settings: 15Mbit | 4Mbit | besteffort bridged-ptm ack-filter
Jun 22 09:29:50 kernel: Cpuidle Host Clock divider is disabled

Have you tried?

/jffs/scripts/cake-qos start XXMbit XXMbit "XXXXX-XXX ack-filter"

substitute X for your required settings

also use

/jffs/scripts/cake-qos status

 

[TechTinkerer]

had crappy results with 95%, now set to 90% looks much better

over wireless phone

 

[brummygit]

Have you tried?

/jffs/scripts/cake-qos start XXMbit XXMbit "XXXXX-XXX ack-filter"

substitute X for your required settings

also use

/jffs/scripts/cake-qos status

Yes it starts up fine manually.

I didn't run a status check after the reboot and would unfortunately get lynched now if I reboot during the day with everyone working from home.

 

[TechTinkerer]

Yes it starts up fine manually.

I didn't run a status check after the reboot and would unfortunately get lynched now if I reboot during the day with everyone working from home.

Yeah, tell me about it ninja rebooter...if they go to the toilet you reboot it's the only way

 

[TechTinkerer]

@brummygit it works fine on reboot i just did a sneaky one
v0.0.4 as i checked status after restart with

/jffs/scripts/cake-qos status

 

[brummygit]

@brummygit it works fine on reboot i just did a sneaky one
v0.0.4 as i checked status after restart with

/jffs/scripts/cake-qos status

Hhhmm - there was a significant difference in buffer bloat. I'm running Dual WAN for failover, I wonder if that has anything to do with it. I'll see if I can find a sneaky slot later and try again, ensuring I wait for Dual WAN to finish all of it's checks and confirm that my Primary WAN is up and running before rushing in to any conclusions.

 

[TechTinkerer]

Just did a test just now on pc this time i do wonder if setting to 85% would make any difference @jackiechun

over wireless pc

 

[brummygit]

Managed a reboot. After running stable for 10 mins, as previously, I get bad bufferbloat results and no speed capping on a test

Running the status check gives

hodgetts@RT-AX88U-AA80:/jffs/scripts# /jffs/scripts/cake-qos status
> CakeQOS-Merlin v0.0.4 is running
> Download Status:
qdisc cake 8002: dev ifb9eth0 root refcnt 2 bandwidth 15Mbit besteffort triple-isolate nat wash ingress ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce
> Upload Status:
qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1

After copying and running the command from my services-start

/jffs/scripts/cake-qos start 15Mbit 4Mbit "besteffort bridged-ptm ack-filter"

I get

hodgetts@RT-AX88U-AA80:/jffs/scripts# /jffs/scripts/cake-qos status
> CakeQOS-Merlin v0.0.4 is running
> Download Status:
qdisc cake 8004: dev ifb9eth0 root refcnt 2 bandwidth 15Mbit besteffort triple-isolate nat wash ingress ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce
> Upload Status:
qdisc cake 8003: dev eth0 root refcnt 2 bandwidth 4Mbit besteffort triple-isolate nat nowash ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce

Notice the difference in the Upload results from the status

 

[heysoundude]

I just installed 0.0.3 not even 24hrs ago and there’s an update already?!?
Is this Covid-era productivity? Traction meeting momentum? Righteously wicked goodness? [emoji2370][emoji1303][emoji1305][emoji1591]


Sent from my iPhone using Tapatalk

 

[TechTinkerer]

Managed a reboot. After running stable for 10 mins, as previously, I get bad bufferbloat results and no speed capping on a test

Running the status check gives

hodgetts@RT-AX88U-AA80:/jffs/scripts# /jffs/scripts/cake-qos status
> CakeQOS-Merlin v0.0.4 is running
> Download Status:
qdisc cake 8002: dev ifb9eth0 root refcnt 2 bandwidth 15Mbit besteffort triple-isolate nat wash ingress ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce
> Upload Status:
qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1

After copying and running the command from my services-start

/jffs/scripts/cake-qos start 15Mbit 4Mbit "besteffort bridged-ptm ack-filter"

I get

hodgetts@RT-AX88U-AA80:/jffs/scripts# /jffs/scripts/cake-qos status
> CakeQOS-Merlin v0.0.4 is running
> Download Status:
qdisc cake 8004: dev ifb9eth0 root refcnt 2 bandwidth 15Mbit besteffort triple-isolate nat wash ingress ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce
> Upload Status:
qdisc cake 8003: dev eth0 root refcnt 2 bandwidth 4Mbit besteffort triple-isolate nat nowash ack-filter split-gso rtt 100.0ms ptm overhead 22 no-sce

Notice the difference in the Upload results from the status

I get the same too, I have mine set at 63Mbit\16Mbit, would be even closer to those limits over direct ethernet

 

[Chris0815]

I just installed 0.0.3 not even 24hrs ago and there’s an update already?!?
Is this Covid-era productivity? Traction meeting momentum? Righteously wicked goodness?

Very dynamic! This thread started on Saturday and today (Monday) already 9 pages with a lot of information and feedback! I am really impressed and try to follow!

 

[heysoundude]

Very dynamic! This thread started on Saturday and today (Monday) already 9 pages with a lot of information and feedback! I am really impressed and try to follow!

There’s a previous thread (10+ pages) as well that goes back a year or 2...once an “official build” happened, it was decided a dedicated thread was most appropriate.


Sent from my iPhone using Tapatalk