CakeQOS CakeQOS-Merlin

[askan7]

Fair enough, I guess.

My asus router has no issues with the frequency of ping signals sent, and I would still consider it a matter that should be investigated. I'm also wary of those who use terms such as "bad networking mamba jamba", but concede the author might simply not want to get into great detail about the subject.

How do I interpret packet loss identified by WinMTR

I've been asked to look into a packet loss problem a previous consultant identified at a client. They have two ADSLMax connections, one for data and the other dedicated to VoIP. The packet loss is

serverfault.com

And I'm still not convinced that the "issue" is an actual issue outside of "omg, the graphs look different", but I will concede that fq_codel does appear to perform slightly better on your network.

Mine also does respond to all pings if they are directly to it.
If I leave ping running directly to router with 0.1s interval while I run another to 1.1.1.1 concurrently I don't get any packet loss on direct ping to router, yet it shows in the ping to 1.1.1.1 but doesn't carry to the final destination.

 

[Audionut]

Mine responds to pings irregardless.

 

[askan7]

Mine responds to pings irregardless.

Using ping plotter?

 

[Audionut]

Well.... if ping plotter reported something different to another ping program, that was also be a cause for concern imo. But yes.

edit: uploaded correct screenshot.

And this screenshot you posted earlier shows packet loss all the way to the last hop.

 

[askan7]

Well.... if ping plotter reported something different to another ping program, that was also be a cause for concern imo. But yes.

View attachment 26976

edit: uploaded correct screenshot.

And this screenshot you posted earlier shows packet loss all the way to the last hop.

You right, but that screenshoot I posted was while I was running speedtest and testing qos settings. I don't have packet loss now, even while loaded. (missed the overhead).
Have no idea why i get packet loss on first hops, don't think it matters since if i run ping directly to ethier of them concurrently it doesn't get packet loss. If you have any ideas don't hesitate to share

Just tried the same on another computer, connect to my isp's router instead of the asus and it's same result. If it's faster than 1second it start to packet loss at hop 3 and slowly starts to packet loss at hop 2 and 1. Doesn't get to final hops.

 

[Morris]

Try using task manager to give you ping application real time. Also use a wired connection.

 

[askan7]

Try using task manager to give you ping application real time. Also use a wired connection.

I did, tried connecting directly to the router my isp provided with different pc. It's seems my isp is blocking ICMP traffic, no big deal i guess. Everything gets to destination just fine without any packet loss.
Another funny thing i noticed is that ping connected to isp's router is like 1ms, everything around 1ms higher ping to the internet. On the Asus i get 0.3ms to router, and even tough it's connected to my isp's router/modem that's in bridge mode, ping is lower even to the outside. I guess faster cpu/nat makes a difference.

With fq-codel my network is pretty, only thing that breaks is bittorrent clients, slows everything down...
Cake is better at pretty much everything but idle latency. If that was fixed I would use Cake without a question. No matter what some people here say, +10ms extra constant jitter is a bit much considering there's no load going on.

 

[Morris]

I did, tried connecting directly to the router my isp provided with different pc. It's seems my isp is blocking ICMP traffic, no big deal i guess. Everything gets to destination just fine without any packet loss.
Another funny thing i noticed is that ping connected to isp's router is like 1ms, everything around 1ms higher ping to the internet. On the Asus i get 0.3ms to router, and even tough it's connected to my isp's router/modem that's in bridge mode, ping is lower even to the outside. I guess faster cpu/nat makes a difference.

With fq-codel my network is pretty, only thing that breaks is bittorrent clients, slows everything down...
Cake is better at pretty much everything but idle latency. If that was fixed I would use Cake without a question.

ICMP type 0 is ping. Nothing you have show would happen if ICMP were blocked.

Internet Control Message Protocol - Wikipedia

en.wikipedia.org

.

 

[Kingp1n]

ICMP type 0 is ping. Nothing you have show would happen if ICMP were blocked.

Internet Control Message Protocol - Wikipedia

en.wikipedia.org

.

Whats the best way to use connmon and setting up the "rtt xxxxx" as an example to the additional settings.

When using connmon, my ping ranges from 25-32 ms.

I also use a full time VPN.

Would setting up "rtt 50000" be a good starting point?

 

[Morris]

Whats the best way to use connmon and setting up the "rtt xxxxx" as an example to the additional settings.

When using connmon, my ping ranges from 25-32 ms.

I also use a full time VPN.

Would setting up "rtt 50000" be a good starting point?

VPN is for security and will interfere with QOS. Chose one or the other

 

[Kingp1n]

I wanted to share my new cake settings ...running for the past few days without any issues and full time VPN:

[1] --> Download Speed | [65 Mbit]
[2] --> Upload Speed | [4 Mbit]
[3] --> Queue Priority | [diffserv4]
[4] --> Extra Download Options |[docsis rtt 50000]
[5] --> Extra Upload Options | [docsis ack-filter rtt 50000]

A+ bufferbloat test !!! Thanks for bringing cake to Asus routers!

 

[askan7]

Guys, I made a clip showing the issue with cake while gaming. Again, no other devices connected, connection is not loaded at all.
Watch the graph. This only happens while gaming, wired or wireless doesn't matter. Idle latency without gaming is same cake on/off.

Tried Battlefield V, Warzone same behaviour. Also pinging from different pc while main pc is gaming, same behaviour.

 

[heysoundude]

seems pretty smooth and quick to me, but I'm not holding the controller. (I can't tell what latency you're feeling)
Also, nvm graphs...is it maybe the program monitoring creating the issue? and the screencap program?

 

[Jack Yaz]

Guys, I made a clip showing the issue with cake while gaming. Again, no other devices connected, connection is not loaded at all.
Watch the graph. This only happens while gaming, wired or wireless doesn't matter. Idle latency without gaming is same cake on/off.

Tried Battlefield V, Warzone same behaviour. Also pinging from different pc while main pc is gaming, same behaviour.

at some point you have to consider if your pc simply cannot keep up with the games and network packet processing is taking a backseat to game rendering. what does your cpu and memory usage look like during the game?

 

[askan7]

at some point you have to consider if your pc simply cannot keep up with the games and network packet processing is taking a backseat to game rendering. what does your cpu and memory usage look like during the game?

It's not the pc, I run pingplotter on laptop and if game is open on desktop it does the same thing.
I'm running a 8700k at 4.8ghz + 16gb 3600mhz cl15 ram.

seems pretty smooth and quick to me, but I'm not holding the controller. (I can't tell what latency you're feeling)
Also, nvm graphs...is it maybe the program monitoring creating the issue? and the screencap program?

yes movement is smooth, extra 10ms jitter doesn't affect movement at all. Also I use mouse not controller, makes me sad to see people playing shooters on pc with controller.
It's not the program either, a simple ping -t shows the issue.

 

[TonyK132]

I have cake set and going on my ac86u with Merlin 384.19 using diffserv4. But I noticed that when I'm watching Netflix, that all traffic seems to be directed to the Bulk category and almost none to the Video category. Is there some configuration that I need to do to get those streams to be correctly handled by cake? The other settings that I have are:

download: unlimited
upload: unlimited
down options: raw nowash rtt 50ms
up options: raw rtt 50ms

Here are the statistics I have:


qdisc cake 8005: root refcnt 2 bandwidth unlimited diffserv4 triple-isolate nat nowash ingress no-ack-filter split-gso rtt 50.0ms raw overhead 0 no-sce
Sent 14243287460 bytes 11145380 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 2Kb of 15140Kb
capacity estimate: 0bit
min/max network layer size: 60 / 1514
min/max overhead-adjusted size: 60 / 1514
average network hdr offset: 14

Bulk Best Effort Video Voice
thresh 0bit 0bit 0bit 0bit
target 2.5ms 2.5ms 2.5ms 2.5ms
interval 50.0ms 50.0ms 50.0ms 50.0ms
pk_delay 17us 14us 16us 7us
av_delay 5us 5us 1us 3us
sp_delay 3us 4us 1us 1us
backlog 0b 0b 0b 0b
pkts 11108726 36256 30 368
bytes 14236703244 6560326 1800 22090
way_inds 354391 386 0 0
way_miss 53393 3558 30 2
way_cols 0 0 0 0
marks 0 0 0 0
drops 0 0 0 0
ack_drop 0 0 0 0
sp_flows 0 1 0 0
bk_flows 1 0 0 0
un_flows 0 0 0 0
max_len 1514 345 60 70
quantum 1514 1514 1514 1514

 

[heysoundude]

I have cake set and going on my ac86u with Merlin 384.19 using diffserv4. But I noticed that when I'm watching Netflix, that all traffic seems to be directed to the Bulk category and almost none to the Video category. Is there some configuration that I need to do to get those streams to be correctly handled by cake? The other settings that I have are:

download: unlimited
upload: unlimited
down options: raw nowash rtt 50ms
up options: raw rtt 50ms

Here are the statistics I have:


qdisc cake 8005: root refcnt 2 bandwidth unlimited diffserv4 triple-isolate nat nowash ingress no-ack-filter split-gso rtt 50.0ms raw overhead 0 no-sce
Sent 14243287460 bytes 11145380 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
memory used: 2Kb of 15140Kb
capacity estimate: 0bit
min/max network layer size: 60 / 1514
min/max overhead-adjusted size: 60 / 1514
average network hdr offset: 14

Bulk Best Effort Video Voice
thresh 0bit 0bit 0bit 0bit
target 2.5ms 2.5ms 2.5ms 2.5ms
interval 50.0ms 50.0ms 50.0ms 50.0ms
pk_delay 17us 14us 16us 7us
av_delay 5us 5us 1us 3us
sp_delay 3us 4us 1us 1us
backlog 0b 0b 0b 0b
pkts 11108726 36256 30 368
bytes 14236703244 6560326 1800 22090
way_inds 354391 386 0 0
way_miss 53393 3558 30 2
way_cols 0 0 0 0
marks 0 0 0 0
drops 0 0 0 0
ack_drop 0 0 0 0
sp_flows 0 1 0 0
bk_flows 1 0 0 0
un_flows 0 0 0 0
max_len 1514 345 60 70
quantum 1514 1514 1514 1514

wow...
If I recall correctly, with pk_ av_ and sp_delay down in the usec (microsecond) range, well below the target of 2.5 milliseconds, you could probably tighten up your rtt setting to at least regional (30ms) if not metro (10ms) and do away with diffserv4 entirely - the sorting takes time - and still keep your bufferbloat at A+

 

[TonyK132]

OK, I'll try 30ms. But I'd still like the QOS categories to work properly. How can I make that happen?

BTW, where can I go to test buffer bloat to see my score (A, A+, etc)?

 

[Treadler]

OK, I'll try 30ms. But I'd still like the QOS categories to work properly. How can I make that happen?

BTW, where can I go to test buffer bloat to see my score (A, A+, etc)?

Speed test - how fast is your internet? | DSLReports, ISP Information

Use our NEW speed test tool to test how fast your broadband or mobile internet connection really is. Read broadband news, information and join our community

www.dslreports.com

 

[TonyK132]

Thanks, but I'm disappointed to see a C for BufferBloat.