FlexQoS FlexQoS 1.2.4 - Flexible QoS Enhancement Script for Adaptive QoS

[dave14305]

There were some users who complained of dropped connections when the firewall restarted with conntrack flushing enabled. It came to a head when @Jack Yaz integrated AutoBW into spdMerlin with a qos restart potentially every 30-60 minutes (can't remember). Seemed like a simple solution at the time, but now we see how it impacts VPN clients.

So I'll have to revert to requiring a firewall restart when FlexQoS applies settings. Jack will probably want to do service "restart_qos;restart_firewall" regardless of whether FlexQoS is installed or not.

I would need to label that option "Do you want your VPN Client to work properly? o Yes o No".

Someone will need to be unhappy. The people who didn't like their connections dropping can disable Conntrack flushing but then be sad that their pre-existing traffic isn't reflecting correctly in the graphs. Or they can have accurate graphs and deal with connections dropping. spdMerlin users may have more frequent complaints if they also use FlexQoS w/ conntrack flushing. I'll work it out with Jack.

Following the last complaint about spdMerlin restarting things too often a threshold option was put in place to only change QoS values if it was exceeded. I seem to remember that I needed to make it clear the threshold was applied to an average of the last 10 results and I was going to consider making the number of tests to average configurable. (note to self: stop writing new scripts the maintenance on the existing ones is very time consuming!)

Moving this conversation back to the FlexQoS thread for further discussion.

 

[cplay]

So I'll have to revert to requiring a firewall restart when FlexQoS applies settings. Jack will probably want to do service "restart_qos;restart_firewall" regardless of whether FlexQoS is installed or not.

I would need to label that option "Do you want your VPN Client to work properly? o Yes o No".

Someone will need to be unhappy. The people who didn't like their connections dropping can disable Conntrack flushing but then be sad that their pre-existing traffic isn't reflecting correctly in the graphs. Or they can have accurate graphs and deal with connections dropping. spdMerlin users may have more frequent complaints if they also use FlexQoS w/ conntrack flushing. I'll work it out with Jack.

Tbh that toggle button for enable firewall restart with flexqos restarting does make the most sense, in terms of not causing non vpn users connections to drop and allowing vpn users to include the firewall restart command when flexqos flushes knowing full well that connections will drop but qos functionality will work?

something like: enable firewall restart whenever flexqos restarts

tooltip: VPN users will need this enabled to ensure bandwidth manual limits work in adaptive qos after saving settings in flexqos.

Will result in dropped connections upon firewall restarting.

Best of both worlds, maybe?

The same option could be added to spdmerlin autobw settings?

 

[cplay]

I don't think it was ever intended to work with Adaptive QoS (only Traditional QoS). But I'm not 100% certain.

Who would know if the Bandwidth priority monitor settings actually apply to adaptive qos if even someone like you doesn’t (considering you dev flexqos).

Maybe @RMerlin could shed light on this?

 

[dave14305]

Tbh that toggle button for enable firewall restart with flexqos restarting does make the most sense, in terms of not causing non vpn users connections to drop and allowing vpn users to include the firewall restart command when flexqos flushes knowing full well that connections will drop but qos functionality will work?

something like: enable firewall restart whenever flexqos restarts

tooltip: VPN users will need this enabled to ensure bandwidth manual limits work in adaptive qos after saving settings in flexqos.

Will result in dropped connections upon firewall restarting.

Best of both worlds, maybe?

The same option could be added to spdmerlin autobw settings?

Always restarting the firewall is the correct thing to do. This is what stock and Merlin do, and what the old FreshJR_QOS script used to do.

Who would know if the Bandwidth priority monitor settings actually apply to adaptive qos if even someone like you doesn’t (considering you dev flexqos).

Maybe @RMerlin could shed light on this?

Settings priorities on the Bandwidth Monitor page updates the traditional qos nvram variable qos_rulelist. There's no visible iptables or tc commands issued to set those priorities under Adaptive QoS. I just don't know if it's by design or a bug since it's closed source components.

 

[cplay]

Always restarting the firewall is the correct thing to do. This is what stock and Merlin do, and what the old FreshJR_QOS script used to do.

Settings priorities on the Bandwidth Monitor page updates the traditional qos nvram variable qos_rulelist. There's no visible iptables or tc commands issued to set those priorities under Adaptive QoS. I just don't know if it's by design or a bug since it's closed source components.

restarting the firewall is perfect for me and other flexqos vpn users .

regarding device priorities on the bandwidth monitor page, from your answer was I correct to deduce that setting device priorities on bandwidth monitor page ONLY affects traditional qos not adaptive qos as it literally has no mechanism to communicate with adaptive qos device and traffic management, right?

 

[dave14305]

I've pushed an update to the develop branch that incorporates the firewall restart again. If you are a spdMerlin user and you use this develop branch with the new update, the FlexQoS customizations won't be re-applied when spdMerlin updates the bandwidth settings and restarts qos. Will wait for Jack to consider his plan on adding the firewall restart in spdMerlin before pushing this release to stable.

Welcome to the wild side of the develop branch!

 

[cplay]

I had forgot to mention this as we’ve been just talking about my small issue:

Your plugin is so much better than default adaptive qos. Asus should be paying you to work on their adaptive qos.

 

[dave14305]

regarding device priorities on the bandwidth monitor page, from your answer was I correct to deduce that setting device priorities on bandwidth monitor page ONLY affects traditional qos not adaptive qos as it literally has no mechanism to communicate with adaptive qos device and traffic management, right?

It has a mechanism inside the closed-source components if it wanted to use it. We just don't know if it ever planned to use it or not. It's an ASUS/Trend Micro question.

 

[dave14305]

I had forgot to mention this as we’ve been just talking about my small issue:

Your plugin is so much better than default adaptive qos. Asus should be paying you to work on their adaptive qos.

They already pay Trend Micro to work on it. My efforts are antagonistic to their design. But I hope they don't mind.

 

[cplay]

I've pushed an update to the develop branch that incorporates the firewall restart again. If you are a spdMerlin user and you use this develop branch with the new update, the FlexQoS customizations won't be re-applied when spdMerlin updates the bandwidth settings and restarts qos. Will wait for Jack to consider his plan on adding the firewall restart in spdMerlin before pushing this release to stable.

Welcome to the wild side of the develop branch!

thanks for that!

where can I find details on the dev branch of flexqos (changelog)/ where to download it?

is it available to download and install via amtm?

I’ve heard good things about your new fq codel implementation, what have you changed with it?

 

[cplay]

It has a mechanism inside the closed-source components if it wanted to use it. We just don't know if it ever planned to use it or not. It's an ASUS/Trend Micro question.

got it. So just to be clear, currently device priority settings on bandwidth monitor do not affect adaptive qos in anyway currently, correct?

 

[Treadler]

They already pay Trend Micro to work on it. My efforts are antagonistic to their design. But I hope they don't mind.

Very happy with the current ‘develop’ branch, it ‘just works’.
A question, is it possible for an IPv6 address to be allowed as the remote IP address in the iptables rules customisation area?

 

[dave14305]

thanks for that!

where can I find details on the dev branch of flexqos (changelog)/ where to download it?

is it available to download and install via amtm?

I’ve heard good things about your new fq codel implementation, what have you changed with it?

See this post.

FlexQoS - FlexQoS 1.2.4 - Flexible QoS Enhancement Script for Adaptive QoS

BETA Version 1.1.1 (develop branch) NEW Experimental htb+fq_codel qdisc to replace htb+htb+sfq (opt-in) HTB burst parameter now is based on "burst by duration" concept borrowed from SQM scripts @ OpenWRT. Burst is now defined as the number of bytes that can be sent at line speed in 1...

www.snbforums.com

 

[dave14305]

Very happy with the current ‘develop’ branch, it ‘just works’.
A question, is it possible for an IPv6 address to be allowed as the remote IP address in the iptables rules customisation area?

Perhaps in the future. I don’t know enough today to understand all the requirements to validate IPv6 input.

 

[BoostOver]

With game in mind.
An idea.
Can you add real-time bandwidth usage in this TAB? (Maybe a colored bar that fills the field)
It would help to understand exactly which ip and port are used ........... this to create right rules in iptables Rules.
Thanks.
Good Job.

 

[ChatmanR]

I just want Emby and Plex to be recognized as Streaming...

 

[DanNuke86]

@dave14305 ive not managed to recreate the issues I had last week. I did amend the ip tables rules I had, so may have been an issue with something I’d previously done.

Ive nearly always got A+,A,A+ on DSL bufferbloat results. Haven’t been able to test if it’s any better on the develop branch. I’d say overall looking at Spdmerlin and conmon, jitter seems to have been reduced overall, up to 50% less at times. It’s hard to nail down whether (and how much) statistically better it is. I did find the previous stable branch very good as well. I’m a complete novice at all this stuff and been gleaming information off this forum for the last year to educate myself and try to improve my home network.

I still find that if my wife is streaming on the tablet (ax client) or on zoom, I notice a drop in latency performance during online multiplayer gaming. I suspect the issue is with my isp though as it’s coax cable broadband and I’m still on docsis3.0...

Many thanks for your work on FlexQoS.

 

[cplay]

See this post.

FlexQoS - FlexQoS 1.2.4 - Flexible QoS Enhancement Script for Adaptive QoS

BETA Version 1.1.1 (develop branch) NEW Experimental htb+fq_codel qdisc to replace htb+htb+sfq (opt-in) HTB burst parameter now is based on "burst by duration" concept borrowed from SQM scripts @ OpenWRT. Burst is now defined as the number of bytes that can be sent at line speed in 1...

www.snbforums.com

Have changed to the develop branch.

I notice a new option for fq_codel now directly in flexqos or default.

I am a little confused as to the difference between selecting fq_codel above vs having fq_codel selected in the adaptive qos asus screen:

​

Queue Discipline	sfq codel fq_codel

I am running 384.18.

On another note, am using your develop branch with the added firewall restart after applying flexqos and bandwidth download limiter is now working with VPN tunnels

Thanks for fixing it so fast.

Last thing (sorry, I am a not versed in this area so I literally need it spelt out to avoid confusion);

dave14305 said:
It has a mechanism inside the closed-source components if it wanted to use it. We just don't know if it ever planned to use it or not. It's an ASUS/Trend Micro question.

got it. So just to be clear, currently device priority settings on bandwidth monitor do not affect adaptive qos in anyway currently, correct?

 

[dave14305]

I am a little confused as to the difference between selecting fq_codel above vs having fq_codel selected in the adaptive qos asus screen

My fq_codel option replaces ASUS' device-level hierarchy of htb classes with a single fq_codel qdisc to manage class-level traffic for all devices in the same qdisc. This doesn't mean a lot for most users, but the idea is that there will be better fairness among devices within a particular QoS class (e.g. Work-From-Home, Streaming, Gaming, etc.). It makes the QoS selection of sfq vs. codel vs. fq_codel irrelevant mostly.

Last thing (sorry, I am a not versed in this area so I literally need it spelt out to avoid confusion);

You keep asking that question and I think I keep saying that I'm not 100% sure. I don't use it and I don't see evidence of it doing anything when I've tested it.

 

[cplay]

My fq_codel option replaces ASUS' device-level hierarchy of htb classes with a single fq_codel qdisc to manage class-level traffic for all devices in the same qdisc. This doesn't mean a lot for most users, but the idea is that there will be better fairness among devices within a particular QoS class (e.g. Work-From-Home, Streaming, Gaming, etc.). It makes the QoS selection of sfq vs. codel vs. fq_codel irrelevant mostly.

You keep asking that question and I think I keep saying that I'm not 100% sure. I don't use it and I don't see evidence of it doing anything when I've tested it.

Perfect - I'll just leave it as QOS ASUS selection fq_codel (the one where I can pick between sfq,codel,and fq_codel) AND I've also now selected fq_codel from directly within flexQOS (the new option available in develop branch).

Yes, I think your version of answering the question with all the associated technical terminology is not received as an answer to my qos layman brain

I do now understand your answer to the question.

Thanks again for adding the fix to the develop branch so quickly.

Have a lovely weekend.