FlexQoS FlexQoS 1.2.4 - Flexible QoS Enhancement Script for Adaptive QoS

[dave14305]

Is something wrong my adaptive QoS? I don't see the "Wan Packet Overhead" option. The other QoS Types have the overhead option.

View attachment 31431

View attachment 31432

I'm running the AX86U on 386.1_2

I apologize if this has been covered already.

In FlexQoS it is only available on the develop branch, not yet released for stable branch.

 

[tmcb82]

Is something wrong my adaptive QoS? I don't see the "Wan Packet Overhead" option. The other QoS Types have the overhead option.

View attachment 31431

View attachment 31432

I'm running the AX86U on 386.1_2

I apologize if this has been covered already.

@dave14305 can be more specific but I can say Adaptive QOS has never had this option. I can assume it's already factored in somehow in the adaptive logic?

 

[abracadabra11]

Are you running with fq_codel enabled? I imagine Google is using the same IPs and port 443 which would make it hard to come up with a rule for it.

Your best option is to limit the upload in the app.

Not running fq_codel. Why would that make a difference (don't understand fq_codel enough to know myself)?

I'll definitely consider limiting in app if there's no way to do it in Flex.

 

[dave14305]

Not running fq_codel. Why would that make a difference (don't understand fq_codel enough to know myself)?

I'll definitely consider limiting in app if there's no way to do it in Flex.

Because with fq_codel, sparse (normal) traffic should be given higher priority than bulk (large) transfers within the same class.

 

[abracadabra11]

Because with fq_codel, sparse (normal) traffic should be given higher priority than bulk (large) transfers within the same class.

Got it - I'll give it a shot.

 

[jerrytouille]

@dave14305 May I ask what your preference is for the Router Outbound Traffic Class? I currently have it set as the default File Downloads.

Thank you.

 

[dave14305]

@dave14305 May I ask what your preference is for the Router Outbound Traffic Class? I currently have it set as the default File Downloads.

Thank you.

Mine is set to the default.

 

[dave14305]

I've pushed another develop branch update that enables a less-than-ideal method to reorder iptables rules in the list. All your existing iptables rules are loaded into the Well-Known dropdown and can be added to the list in any order you choose. But you must first delete the rule you want to add, otherwise you would trigger the duplicate rule check.

So if you wanted to completely reorder your rules, you would delete them all from the list, then use the dropdown to add them back in your preferred order. If you've customized a Pre-defined rule, be sure to choose the rule under User-defined rules or else you'll get the original version (the same rule name can exist in either section of the dropdown). And of course, click Apply afterward to save them.

Please test it out and let me know what bugs you find. I'm sure there are a few.

I've also picked up the expanded overhead changes from 386.2 alpha.

EDIT: almost forgot, I changed the matching of IP fields in the Tracked Connections filter. The fields are still partial-match, but will perform an exact-match if you add $ to the end of the string (e.g. 192.168.50.10$ will only match 50.10, and not 50.100 or 50.104). When picking a Local IP from the dropdown list, this is added automatically.

 

[chris.at]

Just some thoughts regarding the iptables rule reordering - maybe some comment/hint that the last match wins would also be nice (maybe in the title of the section as a remark) because I think someone (at least me ) could forget it when editing after several months. And if I'm remembering right then in appdb rules first match wins? So maybe another comment/hint there would be nice too.

 

[dave14305]

Just some thoughts regarding the iptables rule reordering - maybe some comment/hint that the last match wins would also be nice (maybe in the title of the section as a remark) because I think someone (at least me ) could forget it when editing after several months. And if I'm remembering right then in appdb rules first match wins? So maybe another comment/hint there would be nice too.

AppDB rules don't matter as much. The order is only important within the webui when trying to evaluate wildcard appDB rules versus non-wildcard rules.

I will see what I can come up with for the iptables hint. Thanks!

 

[chris.at]

Just switched to develop again right now and have to ask if your wan packet overhead setting "ethernet vlan" maybe contains a typo with "42", because on traditional qos it ist set to "4".
best wishes, chris

 

[dave14305]

Just switched to develop again right now and have to ask if your wan packet overhead setting "ethernet vlan" maybe contains a typo with "42", because on traditional qos it ist set to "4".
best wishes, chris

I come from the future to tell you the times they are a-changing!

The logic inferred from the cake source (which is why Merlin changed it) is that the cake ethernet keyword adds 38 bytes of overhead, and the ether-vlan keyword adds an extra 4 bytes. 38+4=42. It's all in testing, so it may change.

 

[maghuro]

I've pushed another develop branch update that enables a less-than-ideal method to reorder iptables rules in the list. All your existing iptables rules are loaded into the Well-Known dropdown and can be added to the list in any order you choose. But you must first delete the rule you want to add, otherwise you would trigger the duplicate rule check.

View attachment 31512

So if you wanted to completely reorder your rules, you would delete them all from the list, then use the dropdown to add them back in your preferred order. If you've customized a Pre-defined rule, be sure to choose the rule under User-defined rules or else you'll get the original version (the same rule name can exist in either section of the dropdown). And of course, click Apply afterward to save them.

Please test it out and let me know what bugs you find. I'm sure there are a few.

I've also picked up the expanded overhead changes from 386.2 alpha.

View attachment 31513

EDIT: almost forgot, I changed the matching of IP fields in the Tracked Connections filter. The fields are still partial-match, but will perform an exact-match if you add $ to the end of the string (e.g. 192.168.50.10$ will only match 50.10, and not 50.100 or 50.104). When picking a Local IP from the dropdown list, this is added automatically.

Nice job!

Just a question before I do sh*t.... If I delete a rule, hit apply, and then refresh the page, it won't be loaded to the drop-down, correct? It disappears forever?

I mean, the drop-down is volatile? It is being populated everytime we open the page?

 

[dave14305]

Nice job!

Just a question before I do sh*t.... If I delete a rule, hit apply, and then refresh the page, it won't be loaded to the drop-down, correct? It disappears forever?

I mean, the drop-down is volatile? It is being populated everytime we open the page?

Yes, good point. The dropdown is based on the user rules at the time the page is loaded. If you delete a rule and apply, the deleted rule will no longer be in the dropdown when the page refreshes.

 

[maghuro]

And regarding the new MTU... In my case, which I have fiber, I don't want to use anything (is that correct?)

So do I have to insert 0 or use conservative default? I'm not sure..

 

[dave14305]

And regarding the new MTU... In my case, which I have fiber, I don't want to use anything (is that correct?)

So do I have to insert 0 or use conservative default? I'm not sure..

I've only ever had a cable modem (DOCSIS) so that's all I know. When in doubt, leave it zero. BTW, it's an overhead value, not the MTU (2 different things).

 

[maghuro]

I've only ever had a cable modem (DOCSIS) so that's all I know. When in doubt, leave it zero. BTW, it's an overhead value, not the MTU (2 different things).

Sorry it is... Thought well, wrote wrong

Thanks for all your work. 5*!

 

[jerrytouille]

And regarding the new MTU... In my case, which I have fiber, I don't want to use anything (is that correct?)

So do I have to insert 0 or use conservative default? I'm not sure..

I use 44 ethernet overhead for my fiber connection.
*ref. https://openwrt.org/docs/guide-user/network/traffic-shaping/sqm-details#sqmlink_layer_adaptation_tab

 

[QuikSilver]

@dave14305 Would the new firmware with CAKE built in have any effect those those of us using Flexqos?

 

[QuikSilver]

@dave14305 Would the new firmware with CAKE built in have any effect those those of us using Flexqos?

Answer my own question....If you dont want cake don't choose it.