[Release] FreshJR Adaptive QOS (Improvements / Custom Rules / and Inner workings)

[Agentsicker]

Go to the "Classification" tab under "Adaptive QoS" at the top right you will see "FreshJR Mod" click on it. It will open up the rules, and then just add the IP/CIDR under the gaming rule. Which should be the top rule listed.

Hey, I was similar a couple of days ago. If you are just wanting to specify the PS4, you don't need the CIDR address. Just put the IP of your PS4 into the top left box in the 'Iptables Rules' on the Classicication page of FreshJR's mod.

I suggest he put 192.168.1.xxx/32. Which will apply the rule only to that IP. If he has multiple systems, it can be adjusted. I don't recommend just putting the internal ip of the device only. Unless @FreshJR added it so you only need to input the IP alone.

awesome thanks guys! i was overthinking it by a long shot. I am planning on using 192.168.1.xxx/32 as i dont need a range. Thanks again!

 

[Sinner]

awesome thanks guys! i was overthinking it by a long shot. I am planning on using 192.168.1.xxx/32 as i dont need a range. Thanks again!

im pretty sure fresh said himself someplace if you just need 1 IP u can just enter it without the cldr

 

[Vexira]

Burst determines how many packets are zero-rated before QOS kicks in.

Not throttling the first few packets (few kB at most) allows you to reach max speeds quicker and expierence better responsiveness.

For some users, the defaults bursts are simply too high for their connection so this leads to 50-100ms duration spikes in ping when their connection is initially getting saturated.

I kept the default bursts, but will provide a version with lowered bursts.

Would this be an option to toggle between default and the lower burst values?

I'm also extremely curious about testing how it will affect gaming traffic with Netflix and YouTube streams in hd running at the same time.

 

[FreshJR]

Would this be an option to toggle between default and the lower burst values?

I'm also extremely curious about testing how it will affect gaming traffic with Netflix and YouTube streams in hd running at the same time.

Sure. That can be implemented aswell.

 

[lilstone87]

Sure. That can be implemented aswell.

Awesome. I will likely mess around with it myself as well. Look forward to this next build you put out.

 

[Vexira]

Th

Sure. That can be implemented aswell.

Thank you for your hard work
And thank you in advance.

 

[brummygit]

Hi FreshJR. We discussed an issue a while ago but I’m wondering if things have moved on at all.

My ISP TV service transmits some channels using multicast. This used to show reversed in QoS (upload as download and vice versa) but now (I think following an ASUS update that RMerlin then inherited) the QoS doesn’t see the multicast traffic at all.

Is there any way to get my multicast IP TV traffic included in QoS as it’s a bit hit and miss otherwise.

Many thanks for your hard work


Sent from my iPhone using Tapatalk

 

[Sinner]

Considering checkin out the new script but was curious. Is it possible to reverse the sport/dport in webui? I have one rule that requires it to function.

 

[FreshJR]

@brummygit I can’t really look into it unless I can replicate the traffic flow on my end. Do you have any other examples of whitelisted multicast traffic I can find on the web?

@Sinner ???

 

[Sinner]

@brummygit I can’t really look into it unless I can replicate the traffic flow on my end. Do you have any other examples of whitelisted multicast traffic I can find on the web?

@Sinner ???

Haha im still on yur jan18 model mod'd and happy. Im jus lookin at the pics of the new webgui. Its the dream we talked about forever. Looking at it though i cant tell if i can do that for the one rule.

 

[bimdimbo]

So there's some changes to QOS mentioned in the new Merlin firmware (384.10). Is the script compatible with the new firmware? Are there some changes we need to be aware of before upgrading to the new firmware or should we hold off? Thanks.

 

[FreshJR]

@bimdimbo script has typically remained compatible between all firmware updates.

I wouldn’t worry about it breaking.

@Sinner I still have no idea what you mean by reversing the dport/sport fields?

There are local and remote designations and they work as intended.

 

[RMerlin]

@bimdimbo script has typically remained compatible between all firmware updates.

I wouldn’t worry about it breaking.

Especially as there was no GPL update (except for the RT-AX88U), meaning the QoS code is unchanged.

 

[Vexira]

@FreshJR are the new lower bursts a fixed value or is the calculation change in the way they are determined?

 

[NGRhodes]

@FreshJR are the new lower bursts a fixed value or is the calculation change in the way they are determined?

I like the way SQM scripts for Openwrt work, you set a time you want for burst and then it has a function called get_htb_burst() to calculate the the burst for your bandwidth, with sensible minimum included. There is also a function get_htb_quantum() to check the quantum for HTB is never bigger than the burst.
I think there is some codel/fq_codel tuning functions there too, but not had chance to look.

https://github.com/tohojo/sqm-scripts/blob/master/src/functions.sh

 

[Vexira]

I like the way SQM scripts for Openwrt work, you set a time you want for burst and then it has a function called get_htb_burst() to calculate the the burst for your bandwidth, with sensible minimum included. There is also a function get_htb_quantum() to check the quantum for HTB is never bigger than the burst.
I think there is some codel/fq_codel tuning functions there too, but not had chance to look.

https://github.com/tohojo/sqm-scripts/blob/master/src/functions.sh

I wonder how much of that could be applied to adaptive QoS, just out of curiosity.

 

[Sinner]

@FreshJR well ive got to hand it to ya. Your new script and webgui is quite intuitive and works perfectly. I did manual install on 384.5 so i could see under the hood. I did swap snapchat into voip in both files but otherwise works perfectly out of the box for me even on the older firmware.

*edit*
Im excited to try the reduced burst version. I suspect my network would benefit from that. I was asking about that very think like 6mo ago.

 

[skeal]

@FreshJR My turn for a little help. I'm having issues with Spotify swamping the Web Surfing catagory. I tried a few different things and came up empty handed. I don't want to set Web Surfing less than 100 percent bandwidth allocation. How do I make Spotify go into Downloads? The below setting change does not work, again I don't know what I'm doing.

 

[FreshJR]

I like the way SQM scripts for Openwrt work, you set a time you want for burst and then it has a function called get_htb_burst() to calculate the the burst for your bandwidth, with sensible minimum included. There is also a function get_htb_quantum() to check the quantum for HTB is never bigger than the burst.
I think there is some codel/fq_codel tuning functions there too, but not had chance to look.

https://github.com/tohojo/sqm-scripts/blob/master/src/functions.sh

Settings burst as a function of time sounds great. I feel like a simpler approach will be more ideal to avoid endless questions.

I was instead thinking something simpler would be better in where you have two options to (1)change burst to 1MTU or (2)half of what is currently set. Of course sanity checking quantum so doesn’t it swollow more than 1burst per tic.

So users can set absolute minumum bursts for categories that don’t need responsiveness and halve other bursts if they are overall set too high.

(This should work well since the existing stucture already sets bursts as a function of speed. That function might even have a time constant used behind the scenes aswell do all we would be doing in halving that constant).

@skeal
If spotify was identified it would NOT be in web surfing catogory. Spotify’s mark was correctly placed into the streaming category. Your rule would move it away from streaming and into file downloads.

 

[skeal]

If spotify was identified it wouldn’t be in the streaming category. That rule would move it away from streaming and into file downloads.

Do you know the "mark" value for Spotify? Nothing I have tried moves the Spotify traffic away from Web Surfing and into Downloads. What am I doing wrong? I know the remote ports the IP address range I cannot get it to work. Please help me my friend or anyone that has had this issue.

Important Update: I got the port redirection to work. But the traffic now going into Downloads still interrupts my Streaming in spite of giving Streaming all the bandwidth. Is the only solution to reduce the available bandwidth to Downloads? Wouldn't that interfere with speed test results?