RT-AX58U FQ_Codel issues

[LongRangeSkeet]

Hi everyone,

I just picked up this router today and set-it all up and enabled the FQ_Codel in the Adaptive QoS settings page.

I'm provisioned at 200/10 by my ISP and Speedtest.net/Fast.com usually shows it hovering around 230+/12+ with zero % packet loss. Bufferbloat is F.

​

The QoS options seemed so simple but would someone be able to double-check if they are correctly set?

​

I've noticed that with QoS enabled, now I'm getting significantly slower speeds and packet loss in the range of .5-1.6%. Is it normal to take such a hit in speed and packet loss with this setting enabled? Bufferbloat is now B, but can the score be improved? Is there a setting that I've misconfigured or is this just the best this router's hardware capable of doing?

​

Thanks to whomever in advance for your help. I really appreciate it.

​

 

[Jgee]

Hi everyone,

I just picked up this router today and set-it all up and enabled the FQ_Codel in the Adaptive QoS settings page.

I'm provisioned at 200/10 by my ISP and Speedtest.net/Fast.com usually shows it hovering around 230+/12+ with zero % packet loss. Bufferbloat is F.

View attachment 23705 View attachment 23707​

The QoS options seemed so simple but would someone be able to double-check if they are correctly set?

View attachment 23704​

I've noticed that with QoS enabled, now I'm getting significantly slower speeds and packet loss in the range of .5-1.6%. Is it normal to take such a hit in speed and packet loss with this setting enabled? Bufferbloat is now B, but can the score be improved? Is there a setting that I've misconfigured or is this just the best this router's hardware capable of doing?

View attachment 23706 View attachment 23708
​

Thanks to whomever in advance for your help. I really appreciate it.

​

On merlins latest or beta, I can't get adaptive qos to limit speeds. Traditional and bandwidth limiter works fine. On latest official firmware, I can get the adaptive qos to limit download speeds, but not upload. Traditional qos and bandwidth limiter work fine. There are errors related to qos in the logs on either firmware. I do believe they're aware of the issues.

 

[L&LD]

Your maximum speeds with QoS enabled will be lower. I would not use a mobile handheld device to verify the settings you're testing.

Use an Ethernet, wired desktop computer or an AC powered laptop with Performance battery settings instead.

The numbers you should be using are 75% to 95% of your consistent speeds. I would ignore the WAN packet overhead parameter and leave it at zero.

https://www.snbforums.com/threads/q...ct-for-wan-packet-overhead.45392/#post-480689

I would also be using FreshJR QOS script too with your ISP speeds.

https://www.snbforums.com/threads/r...ements-custom-rules-and-inner-workings.36836/

And with amtm installed if you are using RMerlin 384.15_0 or later, you can install FreshJR QOS with a menu too.

 

[LongRangeSkeet]

Your maximum speeds with QoS enabled will be lower. I would not use a mobile handheld device to verify the settings you're testing.

Use an Ethernet, wired desktop computer or an AC powered laptop with Performance battery settings instead.

The numbers you should be using are 75% to 95% of your consistent speeds. I would ignore the WAN packet overhead parameter and leave it at zero.

https://www.snbforums.com/threads/q...ct-for-wan-packet-overhead.45392/#post-480689

I would also be using FreshJR QOS script too with your ISP speeds.

https://www.snbforums.com/threads/r...ements-custom-rules-and-inner-workings.36836/

And with amtm installed if you are using RMerlin 384.15_0 or later, you can install FreshJR QOS with a menu too.

Thanks so much for the tips! I adjusted numbers closer to the actual speed tests @ 227/11 and set the discipline to SQF. Immediately my speed and bufferbloat tests improved.

​

I'll try tinker with the 75%-95% option and packet overhead later tonight.

I've owned ASUS routers before, but I'm new to Merlins. I couldn't find a setting anywhere where I can prioritize traffic from specific devices. Is that something that I can find in this FreshJr QOS? Is it easy to install?

Thanks again.

 

[ColinTaylor]

You are using Traditional QoS which will potentially impact your maximum throughput. Use Adaptive QoS instead.

 

[LongRangeSkeet]

You are using Traditional QoS which will potentially impact your maximum throughput. Use Adaptive QoS instead.

I think that I might be going crazy or something is wrong, but now it has zero impact on bufferbloat no mater which QoS Type or Discipline I now choose. It's permanently stuck at F. At some point I even lost access to internet until I disabled the QoS completely.

If other people are having issues with getting the QoS to work then is this feature buggy at the moment?

 

[LongRangeSkeet]

I've tried every possible permutation in the QoS settings and I'm still struggling to get a good AQM going.

The FQ_codEL doesn't really work at all in both traditional and adaptive to make a difference in the bufferbloat tests. Consistently fails to get anything more than an F.

SFQ doesn't work in adaptive at all but occasionally works with traditional. The tests are all over the place. Anything between an F to an A. Basically random grade everytime a buffer bloat test is run.

Even though my previous XR500 was overall janky, it did however, have an amazing consistent A in the buffer bloat tests.

Any one knows when QoS might get fixed for the RT-AX58U.

Thanks

 

[RMerlin]

Any one knows when QoS might get fixed for the RT-AX58U.

Seems fixed with 384_8563 (or at least traffic does get properly assigned to packet queues in my basic tests).

 

[LongRangeSkeet]

Is that an upcoming version?

I have the latest 384.17 loaded. I've even factory reset the router after flashing over the factory software.

Thanks Merlin

 

[Jgee]

here are some of the errors in my log
QoS: qos rule is less than 22
May 23 21:15:27 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:28 A.QoS: qos_count=0, qos_check=1
May 23 21:15:31 A.QoS: qos rule is less than 22
May 23 21:15:31 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:32 A.QoS: qos_count=1, qos_check=1
May 23 21:15:36 A.QoS: qos rule is less than 22
May 23 21:15:36 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:37 A.QoS: qos_count=2, qos_check=1
May 23 21:15:40 A.QoS: qos rule is less than 22
May 23 21:15:40 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:16:45 kernel: htb: htb qdisc 10: is non-work-conserving?
May 23 21:16:52 kernel: htb: htb qdisc 12: is non-work-conserving?
May 23 21:35:02 kernel: htb: htb qdisc 10: is non-work-conserving?
May 23 21:35:02 kernel: htb: too many events!
May 23 23:11:14 kernel: htb: htb qdisc 13: is non-work-conserving?
May 23 23:11:31 kernel: htb: htb qdisc 12: is non-work-conserving?
May 23 23:15:03 kernel: htb: htb qdisc 15: is non-work-conserving?
May 23 23:59:12 kernel: htb: htb qdisc 17: is non-work-conserving?

 

[Theliel]

Hi all,

Im using alpha Ax58u 384.18. QoS seen work, but like @Jgee said, bandwidh limit dont work, no matter what limit you put on it, so bufferbloat is "normal" without any limit

 

[LongRangeSkeet]

here are some of the errors in my log
QoS: qos rule is less than 22
May 23 21:15:27 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:28 A.QoS: qos_count=0, qos_check=1
May 23 21:15:31 A.QoS: qos rule is less than 22
May 23 21:15:31 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:32 A.QoS: qos_count=1, qos_check=1
May 23 21:15:36 A.QoS: qos rule is less than 22
May 23 21:15:36 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:15:37 A.QoS: qos_count=2, qos_check=1
May 23 21:15:40 A.QoS: qos rule is less than 22
May 23 21:15:40 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
May 23 21:16:45 kernel: htb: htb qdisc 10: is non-work-conserving?
May 23 21:16:52 kernel: htb: htb qdisc 12: is non-work-conserving?
May 23 21:35:02 kernel: htb: htb qdisc 10: is non-work-conserving?
May 23 21:35:02 kernel: htb: too many events!
May 23 23:11:14 kernel: htb: htb qdisc 13: is non-work-conserving?
May 23 23:11:31 kernel: htb: htb qdisc 12: is non-work-conserving?
May 23 23:15:03 kernel: htb: htb qdisc 15: is non-work-conserving?
May 23 23:59:12 kernel: htb: htb qdisc 17: is non-work-conserving?

Damn that's a real bummer. I replaced my XR500 under the impression that Merlin firmware had a functional AQM.

Well hopefully with a future update this will be fixed. Fingers crossed.

That's the one huge advantage of buying an Asus over Netgear.

 

[Jgee]

Damn that's a real bummer. I replaced my XR500 under the impression that Merlin firmware had a functional AQM.

Well hopefully with a future update this will be fixed. Fingers crossed.

That's the one huge advantage of buying an Asus over Netgear.

Haha! Must be a trend. I'm coming off an xr500 after over a year a failed promises. And that's why I came to Asus. If there's a problem, it won't be over a year before it's fixed.

 

[LongRangeSkeet]

Haha! Must be a trend. I'm coming off an xr500 after over a year a failed promises. And that's why I came to Asus. If there's a problem, it won't be over a year before it's fixed.

Sheeeeeiiit. I had mine since August 2018. They literally fixed absolutely nothing since. Soooo many missing and broken features that it's unbelievable that people are still holding onto them. XR500 still feels like an alpha release to this day.

Luckily I bought a 2 year warranty on that $300 POS.

 

[Jgee]

I bought mine as soon as it came out. Sold it, only to have to buy it back from one pissed off customer. Found out the customer ran it without the antennas, as he didn't use it for wireless. Fried the Wi-Fi. I had no better joy than beating that pos to death with its own power brick, then throwing off a second story deck!

 

[LongRangeSkeet]

Is there a way to maybe get CAKE on Merlin through ssh or is QoS totally broken at the moment?

Is there at least a way to prioritize device traffic for now?

 

[Jgee]

Is there a way to maybe get CAKE on Merlin through ssh or is QoS totally broken at the moment?

Is there at least a way to prioritize device traffic for now?

Traditional qos works for me. I know you can prioritize by device or ports. Limits work, so no bufferbloat.

 

[RMerlin]

Is there a way to maybe get CAKE on Merlin through ssh or is QoS totally broken at the moment?

Is there at least a way to prioritize device traffic for now?

Getting another queue scheduler added won't do anything, you'd have to manually configure all your traffic queues yourself since Adaptive QoS has no way of working with that. The only reason I'm able to do it with fq_codel is because it can be used more or less as a drop-in replacement for sfq.

 

[Jgee]

For now, I've had the best results with the bandwidth limiter. One rule. 192.168.1.2-254 set my speeds to 5% lower. Never have a problem with lag, buffering or FaceTime and such. Triple a+ on dslreports, even though everyone says those results don't matter. Four adults one kid. At least 3 streams going all the time. Two heavy gamers. All on 75/15 speeds.

 

[LongRangeSkeet]

Getting another queue scheduler added won't do anything, you'd have to manually configure all your traffic queues yourself since Adaptive QoS has no way of working with that. The only reason I'm able to do it with fq_codel is because it can be used more or less as a drop-in replacement for sfq.

I just tested all three queue disciplines : sfq, codel, and fq_codel. I can now confirm without a reason of a doubt that none of them work.

I tested all three disciplines in Traditional QoS because if set to Adaptive QoS, the dslreports bufferbloat test is always F regardless of the chosen queue discipline.

I've even inputted 70% of my download/upload bandwidth.

I tested with the server statistics graph in Battlefield 1 in an empty server. I ran fast.com to saturate my bandwidth.

Normally, the ping to server is a stable 47ms at all times. During a fast.com test, the ping would spike up to 200+ms.

Hate to say it but the QoS, at least in the RT-AX58U, is completely broken....