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

[deddc23efb]

I do have a lot of tc rules, maybe that's the problem?

@RT-AC3100-0000:/jffs/scripts# tc filter show dev br0
filter parent 1: protocol all pref 1 u32
filter parent 1: protocol all pref 1 u32 fh 800: ht divisor 1
filter parent 1: protocol all pref 1 u32 fh 800::800 order 2048 key ht 800 bkt 0 flowid 1:2
  mark 0x0000 0xc0000000 (success 350927)
filter parent 1: protocol all pref 2 u32
filter parent 1: protocol all pref 2 u32 fh 827: ht divisor 1
filter parent 1: protocol all pref 2 u32 fh 827::800 order 2048 key ht 827 bkt 0 flowid 1:17
  mark 0x80000000 0xc000ffff (success 1182641)

filter parent 1: protocol all pref 10 u32
filter parent 1: protocol all pref 10 u32 fh 81f: ht divisor 1
filter parent 1: protocol all pref 10 u32 fh 81f::800 order 2048 key ht 81f bkt 0 flowid 1:15
  mark 0x80070000 0xc03f0000 (success 0)

Well, the stats are definitely showing that your Default (1:17) flowId is getting hit a lot. There is nothing going to Other (1:15).
Given that you don't see the syslog entry, I still think the script isn't running.
You can invoke it manually ... simply run "FreshJR_QOS". Perhaps you're hitting something annoying like Dos formatting messing up the script.
Line 50 in FreshJR's version of the script has a log event that should show up in your syslog. (If you want you can change it to "logger -s " to have it output to stdout as well.)
If you do not see that log event, the script isn't running.

Other evidence of this is that you don't see custom rules that FreshJR put in the script ... they go at Pref 2 as well.

In my personal version I've added a log event at the end as well just so I know it has completed and done something.

 

[Nodiaque]

hmmm... yeah something is def wrong, it's like the script isn't runable:

 

[Nodiaque]

I deleted the file and copy/paste the whole code directly into putty. Now I'm getting error on runtime, which is because of the bad paste done by the copy in nano. I'll reupload the file

 

[Nodiaque]

That's weird, I'm getting the same error, script not found, when I upload the file again

 

[Nodiaque]

Ok, I decided to make a copy/paster from the old freshjr 192 script, now it seems to compile something, I'm waiting for it to end

 

[deddc23efb]

Someone else had a similar problem. Do a dos2unix on the file - you can do this in place in the cli.

 

[Nodiaque]

Bah, I replaced the file from the old freshjr that was doing something, now I have a log entry stating it has started the modification. I'm still waiting on the prompt return. Is that suppose to be long?

 

[deddc23efb]

Bah, I replaced the file from the old freshjr that was doing something, now I have a log entry stating it has started the modification. I'm still waiting on the prompt return. Is that suppose to be long?

Yeah it has a long wait (3 minutes) to avoid colliding with ASUS filter startup.

 

[Nodiaque]

Ah, ok. Yeah it just returned and I got some loging information. It does seems to work right now, I'll have an eye on it. Thank you!

 

[deddc23efb]

Good news.

Ah, ok. Yeah it just returned and I got some loging information. It does seems to work right now, I'll have an eye on it. Thank you!

Good news.

You can double check that the config is correct by examining the pref 2 filters in your tc filter show dev br0 output (and eth0). You should see that the filter that has the 0x8000ffff goes to in your case I think flowid 1:15 rather than flowid 1:17 .

 

[medwatt]

Hi medwatt,
FreshJR's script doesn't map anything to Net Control. It mostly moves traffic from Default to whatever is set for Other. What variant of 382 are you using? I run his script successfully on 382_1.

Did you run the new script on a clean reboot or did it try to fix changes from a previous run of the old script (it reads data from the current config to do its work.) It shouldn't really matter but perhaps if you're running 382_2 or later there is something the script doesn't expect.

If you are comfortable with using "tc" in CLI you can examine the filter changes that it has made to diagnose.
For example:
show current Download filters in order. The ones at the top of the list (pref 1 , pref 2) are where the script makes its changes.
tc filter show dev br0
show hit counts for traffic filters. The non-zero success ones will tell you what filters are being applied to your traffic:
tc filter show dev br0 | grep -B1 "success [1-9]"

/dedd

Thanks for your reply. I'm currently on Merlin 382_2. All my downloads get categorized either as "Web Surfing" or "Net Control Packets" and as a result they talk the highest priority.

https://pastebin.com/CueyUxP4

 

[deddc23efb]

Thanks for your reply. I'm currently on Merlin 382_2. All my downloads get categorized either as "Web Surfing" or "Net Control Packets" and as a result they talk the highest priority.

https://pastebin.com/CueyUxP4

medwatt if you include the -B1 flag on your grep it will show you 1 line before the match. That tells you what treatment the traffic is going to (the flow Id at the end of the line.) An efficient version of the command that only shows non zero hits and includes the flow id line would be "
tc filter show dev br0 | grep -B1 "success [1-9]"

edited... kept screwing up the end of the command..

 

[medwatt]

medwatt if you include the -B1 flag on your grep it will show you 1 line before the match. That tells you what treatment the traffic is going to (the flow Id at the end of the line.) An efficient version of the command that only shows non zero hits and includes the flow id line would be "
tc filter show dev br0 | grep -B1 "success [1-9]"

edited... kept screwing up the end of the command..

Here you go :

https://pastebin.com/27amyPwQ

 

[deddc23efb]

Hmm. Yes. The entries you highlighted originally are being classed as category 0 and category 1 traffic. The last digit in the flowid is the "traffic priority" 0 highest, 7 lowest (original default).

I don't see anything really wrong in your output.

It looks like your filters are correct ... or at least the same as on my router. The Net Control filter hit is matching the packets as marked by ASUS and putting them in priority 0. The question for you is WHY is ASUS tagging them that way - are you certain that is incorrect? Have you updated your trend micro signatures? Have you you configured IP tables rules that could be remarking these packets for processing by tc?

How are you deciding that the traffic is being classified incorrectly? I watch the pps in QoS Statistics screen to what is instantaneously active.

Wish I could be more help.

/dedd

 

[medwatt]

Hmm. Yes. The entries you highlighted originally are being classed as category 0 and category 1 traffic. The last digit in the flowid is the "traffic priority" 0 highest, 7 lowest (original default).

I don't see anything really wrong in your output.

It looks like your filters are correct ... or at least the same as on my router. The Net Control filter hit is matching the packets as marked by ASUS and putting them in priority 0. The question for you is WHY is ASUS tagging them that way - are you certain that is incorrect? Have you updated your trend micro signatures? Have you you configured IP tables rules that could be remarking these packets for processing by tc?

How are you deciding that the traffic is being classified incorrectly? I watch the pps in QoS Statistics screen to what is instantaneously active.

Wish I could be more help.

/dedd

I am downloading files using internet download manager and the download manager on my web browser. In fact, no traffic gets directed into the "File Transfer" container. Presently, the internet is unusable when someone is downloading something. I'm going to return back to the previous firmware.

 

[deddc23efb]

I am downloading files using internet download manager and the download manager on my web browser. In fact, no traffic gets directed into the "File Transfer" container. Presently, the internet is unusable when someone is downloading something. I'm going to return back to the previous firmware.

Release 382.1_2 works well with the script FreshJR published (make sure the file has no odd Dos file problems ... dos2unix). What release were you testing?

/dedd

 

[medwatt]

Release 382.1_2 works well with the script FreshJR published (make sure the file has no odd Dos file problems ... dos2unix). What release were you testing?

/dedd

That firmware version does not exist for my device (AC68U). I'm back to v380.67 where things work fine. I hope FreshJR will take a look at my issue.

 

[deddc23efb]

That firmware version does not exist for my device (AC68U). I'm back to v380.67 where things work fine. I hope FreshJR will take a look at my issue.

I'm not sure much can be done for you by the tool. The issue appears to be the pre-classification of your traffic.

It occurred to me that you can influence that by changing the priority treatment for devices within the GUI. Did you happen to try to prioritize different devices on the QOS-Bandwidth Monitor screen? If you had bumped your test device to high priority, the Trend Micro marking might look .. odd.. just a guess.
Hope it goes well for you.

/Dedd

 

[medwatt]

Release 382.1_2 works well with the script FreshJR published (make sure the file has no odd Dos file problems ... dos2unix). What release were you testing?

/dedd

Can you download this file (https://drive.google.com/uc?id=18CnyqnxlVpLj1qTdt1I5RIhTBi2Uy4xn&export=download) and tell me which category it falls under in QoS statistic page ?

 

[deddc23efb]

Lol - a huge spike in Net Control. Clearly ASUS is misclassifying that traffic. Probably because it's google.