FlexQoS FlexQoS 1.5.2 - Flexible QoS Enhancement Script for Adaptive QoS

[ExtremeFiretop]

I believe this is how FC control works:

Post in thread 'ax88u-pro: need adaptive qos advice for 5 adults on a 300/300 isp' https://www.snbforums.com/threads/a...r-5-adults-on-a-300-300-isp.95307/post-965702

Normally when qos is disabled FC should be Enabled (default).

Interesting, let me look into this and see if I can replicate

 

[ExtremeFiretop]

@dave14305

I see a hard coded list of models here for flow control added in version: v1.4.0 (11-May-2024)
I'm assuming this hard coded list is outdated considering i don't see any further updates since that point.

Can I ask you what requirements determined if a model made it onto the list?
Can we add code to check a nvram value for fc_disable instead of hard coding models?

 

[dave14305]

I'm assuming this hard coded list is outdated considering i don't see any further updates since that point.

Can I ask you what requirements determined if a model made it onto the list?

They were the existing HND5.04 models at the time, presumably. Are there any newer models? We know the BE models don’t work. I must not have had a way to determine the kernel version in the GUI so the models were hardcoded.

Can we add code to check a nvram value for fc_disable instead of hard coding models?

It wouldn’t achieve the same result, unless I misunderstand what you’re going to do with it.

 

[ExtremeFiretop]

They were the existing HND5.04 models at the time, presumably. Are there any newer models? We know the BE models don’t work. I must not have had a way to determine the kernel version in the GUI so the models were hardcoded.

It wouldn’t achieve the same result, unless I misunderstand what you’re going to do with it.

Gotcha, so it's ONLY HND5.04 models and nothing newer essentially.
Will make testing this tricker, but I can probably just spoof the model or add mine to the list temporarily right to test?

 

[ExtremeFiretop]

@dave14305 Nvm I found a reliable way to test.
@Kingp1n I managed to recreate your issue now in testing, I believe I have a good idea where to look. Give me some time this evening when I return to check into this further

 

[ExtremeFiretop]

Release Notes for FlexQoS 1.5.1 production release now available
[2025-Sep-06]

What's changed/new?

• FIXED: Properly compare version by using arithmetic conversion
• FIXED: Fixed new schedule keeping Flow Cache disabled when disabling QoS as reported by @Kingp1n

The fork from @dave14305 's repository is now hosted on the AMTM-OSR GitHub repo:
https://github.com/AMTM-OSR/FlexQoS

 

[ExtremeFiretop]

Apparently FlexQoS has two modes: six-month nap and 24-hour firehose.
Not sure if I fixed the scheduler, or just disabled "Flow Control" on myself triggering a denial-of-sleep attack.

 

[Kingp1n]

Release Notes for FlexQoS 1.5.1 production release now available
[2025-Sep-06]

What's changed/new?

• FIXED: Properly compare version by using arithmetic conversion
• FIXED: Fixed new schedule keeping Flow Cache disabled when disabling QoS as reported by @Kingp1n

The fork from @dave14305 's repository is now hosted on the AMTM-OSR GitHub repo:
https://github.com/AMTM-OSR/FlexQoS

Thanks for the quick fix...all working flawlessly. I like this new feature.

 

[ExtremeFiretop]

Thanks for the quick fix...all working flawlessly. I like this new feature.

I'm trying.. I'm just happy I didn't really have to re-do any fix twice. (so far)

I have a feeling by this point we must of covered more or less anything missed in the "dev" branch. I'll try not to Jinx anything by saying that out loud though I'm happy to stop at only 24 hours of patching.

 

[Squall Leonhart]

This is not a bug with the code, but a limitation within Firefox.

What you call limitation is a to spec implementation of 'input time', anything chrome is doing is beyond the spec and unreliable, the button won't be present on any non chromium engine browser including safari, if you want the dial it should be done in a completely cross browser manner using something like timepicker-ui

 

[ExtremeFiretop]

If you want the dial it should be done in a completely cross browser manner using something like timepicker-ui

No thanks, not interested in adding anything else at this time.

 

[Squall Leonhart]

No thanks, not interested in adding anything else at this time.

i read that as "Thanks, i prefer to not be spec compliant"

 

[ExtremeFiretop]

i read that as "Thanks, i prefer to not be spec compliant"

No, the purpose of the messaging of the limitation was to inform people not to report the missing dial as an issue.

So adding extra code to support a dial, which I am already saying is unsupported feels like potential technical debt that would need to be supported and updated with time, and I'm not interested in handling that at this time.

If you'd like to submit a PR implementing it, and are willing to support it if something breaks then by all means we will consider it.

 

[maghuro]

I rebooted my router while QoS was off via cron.
When it rebooted, QoS failed to start, which is expected.
However, at the ON time, it started borked. It seemed that QoS and flexqos started, but the page wasn't mounted. I made a buffer float test and it seemed it wasn't working properly.
Solution was forcing an update, and then the aos started flawlessly.

 

[ExtremeFiretop]

I rebooted my router while QoS was off via cron.
When it rebooted, QoS failed to start, which is expected.
However, at the ON time, it started borked. It seemed that QoS and flexqos started, but the page wasn't mounted. I made a buffer float test and it seemed it wasn't working properly.
Solution was forcing an update, and then the aos started flawlessly.

Thanks for the feedback and I managed to replicate this on my side.
Essentially if I had a schedule set to disable QoS at noon and it rebooted at 12:30 (for example), it would not actually "startup" FlexQoS since it has a check for QoS being enabled at startup otherwise it skips, and the last state was having QoS disabled before the reboot.

Sep 27 11:58:56 FlexQoS: /jffs/addons/flexqos/flexqos.sh (pid=16848) called in unattended mode with 1 args: -start
Sep 27 11:58:56 FlexQoS: Adaptive QoS is not enabled. Skipping FlexQoS startup.

This caused the bug you were seeing.
I had 2 choices which was remove the check from the startup function, but I believe it's there for good reason. (So that we don't attempt to apply any policies while QoS is disabled.)

My other choice which is what I elected to do was re-ordered and organized the flow for the startup sequence so it always starts QoS if you have a schedule configured.
This allows FlexQoS to startup normally, and then at the end I apply any custom schedules which may (or may not) re-disable QoS depending on the current time.

Please update to the latest develop version to test again (make sure you have QoS enabled when you do so):

sh /jffs/addons/flexqos/flexqos.sh develop

 

[maghuro]

Thanks for the feedback and I managed to replicate this on my side.
Essentially if I had a schedule set to disable QoS at noon and it rebooted at 12:30 (for example), it would not actually "startup" FlexQoS since it has a check for QoS being enabled at startup otherwise it skips, and the last state was having QoS disabled before the reboot.

Sep 27 11:58:56 FlexQoS: /jffs/addons/flexqos/flexqos.sh (pid=16848) called in unattended mode with 1 args: -start
Sep 27 11:58:56 FlexQoS: Adaptive QoS is not enabled. Skipping FlexQoS startup.

This caused the bug you were seeing.
I had 2 choices which was remove the check from the startup function, but I believe it's there for good reason. (So that we don't attempt to apply any policies while QoS is disabled.)

My other choice which is what I elected to do was re-ordered and organized the flow for the startup sequence so it always starts QoS if you have a schedule configured.
This allows FlexQoS to startup normally, and then at the end I apply any custom schedules which may (or may not) re-disable QoS depending on the current time.

Please update to the latest develop version to test again (make sure you have QoS enabled when you do so):

sh /jffs/addons/flexqos/flexqos.sh develop

Nothing to complain on latest develop version

Thanks for your prompt help

 

[ExtremeFiretop]

Nothing to complain on latest develop version

Thanks for your prompt help

Happy to help. If nothing else is reported I'll issue a new release come September 30th.

I'll be a lot slower come October (if able to help at all), as I bought my first house and I'm getting the keys October 1st.

I'm excited, but there's a little work to do and I'll be busy with the move for the month of October.

I'm considering putting an out of office on in my signature or something lol
All that to say if you find any other bugs this is the time!

 

[Kingp1n]

Happy to help. If nothing else is reported I'll issue a new release come September 30th.

I'll be a lot slower come October (if able to help at all), as I bought my first house and I'm getting the keys October 1st.

I'm excited, but there's a little work to do and I'll be busy with the move for the month of October.

I'm considering putting an out of office on in my signature or something lol
All that to say if you find any other bugs this is the time!

Congrats on the new home. Thanks for the quick fix!

 

[ExtremeFiretop]

Release Notes for FlexQoS 1.5.2 production release now available
[2025-Sep-30]

What's changed/new?
- FIXED: A bug with commit: Bug-Fix caused FlexQoS to fail the startup sequence if the router rebooted while a QoS schedule had QoS disabled. As reported by maghuro

The fork from @dave14305 's repository is now hosted on the AMTM-OSR GitHub repo:

GitHub - AMTM-OSR/FlexQoS: FlexQoS - Flexible QoS Enhancement Script for Adaptive QoS on ASUSWRT-Merlin

FlexQoS - Flexible QoS Enhancement Script for Adaptive QoS on ASUSWRT-Merlin - AMTM-OSR/FlexQoS

github.com

I'm out of here for the next month of October for my move, see you all in about a month!