FlexQoS FlexQoS issues with 388.4 HND5.04 models

[dave14305]

I've pushed version 1.4.0 which includes a "flow cache control" option for FlexQoS on HND5.04 routers.

This option can be Off (default), On, or Auto. On will disable flow cache during FlexQoS initialization, allowing iptables rules to work like they used to. Auto will only disable flow cache if:

1. Flow cache is already on.
2. FlexQoS iptables rules are defined.
3. QoS configured bandwidth is 400 Mbps or less.

I also found an interesting way to tie in to the default tooltip help system used by the firmware.

 

[SomeWhereOverTheRainBow]

I've pushed version 1.4.0 which includes a "flow cache control" option for FlexQoS on HND5.04 routers.

This option can be Off (default), On, or Auto. On will disable flow cache during FlexQoS initialization, allowing iptables rules to work like they used to. Auto will only disable flow cache if:

1. Flow cache is already on.
2. FlexQoS iptables rules are defined.
3. QoS configured bandwidth is 400 Mbps or less.

I also found an interesting way to tie in to the default tooltip help system used by the firmware.

I wonder if now flexQoS can play with cake more since you are allowing a feature to turn off flow cache control. I know you are not feeling adventurous anymore on that subject, but IIRC the flow cache might have been your biggest hindrance when experimenting.

 

[langesgnu]

I've pushed version 1.4.0 which includes a "flow cache control" option for FlexQoS on HND5.04 routers.

This option can be Off (default), On, or Auto. On will disable flow cache during FlexQoS initialization, allowing iptables rules to work like they used to. Auto will only disable flow cache if:

1. Flow cache is already on.
2. FlexQoS iptables rules are defined.
3. QoS configured bandwidth is 400 Mbps or less.

I also found an interesting way to tie in to the default tooltip help system used by the firmware.

If I understand the reasoning here correctly, then there is some issue between flow cache and the iptable rules, which anyhow do not seem to work on HND models and are therefore unnecessary.

Prior to the update flow cache was enables and I only had 1 iptable rule for Wifi calling, which I do not need. Now I set the new option to "auto" and deleted my last iptable rule, which however automatically generates again the 4 standard rules. So the 2nd criteria above can never lead to a situation where flow cache is on, as I cannot delete the iptable rules.

Am I missing something or do I need to wait for flow cache to be turned on again after some time?

 

[SomeWhereOverTheRainBow]

If I understand the reasoning here correctly, then there is some issue between flow cache and the iptable rules, which anyhow do not seem to work on HND models and are therefore unnecessary.

Prior to the update flow cache was enables and I only had 1 iptable rule for Wifi calling, which I do not need. Now I set the new option to "auto" and deleted my last iptable rule, which however automatically generates again the 4 standard rules. So the 2nd criteria above can never lead to a situation where flow cache is on, as I cannot delete the iptable rules.

Am I missing something or do I need to wait for flow cache to be turned on again after some time?

If I am not mistaken, I believe the flow cache was causing the traffic to lose its proper traffic markings, hence why the iptable rules were not working. Once flow cache is turned off the traffic loses this unpredictable behavior with the traffic markings, thus why iptables work properly with flow cache turned off.

 

[dave14305]

If I understand the reasoning here correctly, then there is some issue between flow cache and the iptable rules, which anyhow do not seem to work on HND models and are therefore unnecessary.

Prior to the update flow cache was enables and I only had 1 iptable rule for Wifi calling, which I do not need. Now I set the new option to "auto" and deleted my last iptable rule, which however automatically generates again the 4 standard rules. So the 2nd criteria above can never lead to a situation where flow cache is on, as I cannot delete the iptable rules.

Am I missing something or do I need to wait for flow cache to be turned on again after some time?

If you want flow cache enabled, leave the setting Off. The primary focus is to offer a way to disable fc if iptables rules are desired.

 

[BoostOver]

…

I apologize for my bad English
Some time ago you posted a link in which the importance of uploading in multiplayer games was explained.......I can't find your post anymore. Could you help me?

 

[dave14305]

I apologize for my bad English
Some time ago you posted a link in which the importance of uploading in multiplayer games was explained.......I can't find your post anymore. Could you help me?

The original site appears dead, but you can read comments here:

Why you need at least 3Mbps upload to get good game performance with ~1500byte packets: Doing the math

There are many people who try hard to game on slow asymmetric lines such as DSL with 768kbps upload. Because cake and SQM do such a good job, they often think they can just tune it right and they'll get good game performance, but they find that no matter how much they tune things, it just never...

forum.openwrt.org

 

[BoostOver]

…

..

You are kind as always, thank you. I would like to ask you one last question: As for online gaming, is adaptive QoS preferable.....or sqm? Thanks again for your support.

 

[langesgnu]

If you want flow cache enabled, leave the setting Off. The primary focus is to offer a way to disable fc if iptables rules are desired.

That is what I did. Just wanted to mention, that the "auto" option does not seem to have any purpose, for me it is either flow cache and no iptable rules or the other way round.

 

[dave14305]

That is what I did. Just wanted to mention, that the "auto" option does not seem to have any purpose, for me it is either flow cache and no iptable rules or the other way round.

Not being able to completely remove iptables rules is acknowledged. But if your ISP speeds are below 400 Mbit, then it is likely flowcache won’t make a significant difference if auto-disabled.

 

[dave14305]

As for online gaming, is adaptive QoS preferable.....or sqm?

@Kingp1n is the gaming QoS expert.

 

[Kingp1n]

…

You are kind as always, thank you. I would like to ask you one last question: As for online gaming, is adaptive QoS preferable.....or sqm? Thanks again for your support.

Both work well ...what are your paid speeds? This will determine what you should be using. CAKE is a simpler setup....set it & forget it. Just make sure you setup the allocated speeds at 85-95% of your actual speeds. Start lower and adjust as need it.

FlexQoS will require a little bit more setup (nothing crazy).

When doing a latency test thru these available websites, I always seem to see the CAKE results does slightly a better job on reducing latency (numbers are better) but "real-world" is what matters and I would recommend you run both for a week (try CAKE 1st) and do alot of gaming while testing them. I would choose the one setup where you don't see any lag while gaming and other folks are doing other things i.e. surfing web, streaming etc....

I think you'll be surprised that CAKE will do just fine.

 

[BoostOver]

Both work well ...what are your paid speeds?

100/20

 

[Kingp1n]

100/20

Try CAKE and let us know how it's working for you.

 

[Tomo]

I’m a gamer and i can confirm Kingp1n , my speed are 90/14 vdsl2 and cake is the best!

 

[dave14305]

That is what I did. Just wanted to mention, that the "auto" option does not seem to have any purpose, for me it is either flow cache and no iptable rules or the other way round.

There’s a new version 1.4.1 that should allow you to delete your iptables rules now. Please update and see if it behaves as you would expect. Thanks.

 

[Kingp1n]

There’s a new version 1.4.1 that should allow you to delete your iptables rules now. Please update and see if it behaves as you would expect. Thanks.

@dave14305

"Auto" option working flawlessly. We appreciate this update!!!

 

[_BLACK_BuLLeT_]

How can I limit the speed of a specific site (or at least several IP addresses) using FlexQoS?

I’m currently checking the limitation on Speedtest.net, for this I set the Others class for Speedtest.net in the AppDB Redirection Rules (and moved Untracked to another class).

Spoiler: screenshot

Also in Bandwidth I specified a 1% download and upload speed limit for the Others class.

Spoiler: screenshot

After setting up these rules, I saved them using the Apply button. Thus, when I run a test on the Speedtest.net website, this traffic is classified as Others, which can be seen in the graph.

Spoiler: screenshot

But the speed is not limited to 1%; the Speedtest.net site still works at the maximum speeds of my network. What am I doing wrong? Why might this not work and how can I fix it?

I'm using the latest Merlin firmware and have updated FlexQoS. QoS operating mode selected Adaptive QoS

Spoiler: screenshot

 

[dave14305]

But the speed is not limited to 1%; the Speedtest.net site still works at the maximum speeds of my network. What am I doing wrong? Why might this not work and how can I fix it?

You haven’t manually set the QoS Bandwidth Setting. Instead you have it set to Automatic. 1% of unlimited is still unlimited. What is your ISP speed?

 

[langesgnu]

There’s a new version 1.4.1 that should allow you to delete your iptables rules now. Please update and see if it behaves as you would expect. Thanks.

Sorry for the late reply, but I have not been home for some days. Updated now to 1.4.1 and now I can delete all the iptable rules and now "auto" disables flow cache, probably only because it was previously on.

Though I have to say, I do not fully understand the reason here. When there are no iptable rules present, shouldn't it rather be on, even if speeds are on the lower side?