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

[CriticJay]

Short Question:

Should i enable in the Network adapter im my pc Priority & VLAN ?

And should i enable Microsoft-Message Queue-Server in the Windows functions?

All for gaming ?

your question is a general windows networking question, has nothing to do with FreshJR script, let alone Asuswrt or Asus routers even ...

 

[randomName]

My main rig is getting an A+ Buffer bloat while my gaming rig is getting a C. lol, how ? I've rebooted the router several times via software. Nothing is running according to the pie chart. Using DSLReports/speedtest. I've manually assigned an IP to my game rig. Possible corruption?

 

[FreshJR]

Are you maxing out the wireless link rate?

Buffet bloat is when any pathway is conjested. The wifi can get congested if your internet speed is faster than it.

 

[randomName]

Are you maxing out the wireless link rate?

Buffet bloat is when any pathway is conjested. The wifi can get congested if your internet speed is faster than it.

The only thing that was using packets was the wired PC gaming rig afaik, and all of a sudden the game experience changed and things felt slow. I exited the game, waited till there were no packets in or out, ran the test and I got a C. This was not even an hour after it got an A+. I watch the pie chart with all devices enabled on my main rig while i game on the other. I was using Codel. I changed to fq_codel and that seemed to fix it but like I said I got an A+ with Codel about an hour ago. The only thing I can think of that I did was change the automatic router log out, without a reboot. My smart TV seems to connect to the internet without even being turned on, sometimes. It's very odd. Same with my smart phone. Maybe that was it

 

[Gingernut]

my smart TV seems to connect to the internet without even being turned on, sometimes. It's very odd. Same with my smart phone. Maybe that was it

Most smart devices won't be truly turned off unless you physically unplug them from their power outlet ,they just go into a low power state. My Mi box is always sending/receiving data unless I pull the plug.

We're living Orwellian times I'm afraid.

Your ISP was more than likely the cause of your drastic bufferbloat change.

 

[Sptz]

Hi Fresh,

I've noticed since of late, after updating to the new version (I even uninstalled, and re-installed. Also running the latest merlin on my AC68U) I keep getting a lot of these, is it normal? Undf Prio does show 2 which would mean it's working right? But it does look like I'm getting subpar performance as opposed to a few months ago. Without QOS enabled my internet is the same as it's always been.

Apr 22 09:49:59 miniupnpd[566]: HTTP listening on port 36001
Apr 22 09:49:59 miniupnpd[566]: Listening for NAT-PMP/PCP traffic on port 5351
Apr 22 09:50:01 disk_monitor: be idle
Apr 22 09:50:02 qos: Applying codel patch
Apr 22 09:50:02 adaptive QOS: Applying - Iptable Down Rules
Apr 22 09:50:03 adaptive QOS: Applying - Iptable Up   Rules (eth0)
Apr 22 09:50:03 adaptive QOS: TC Modification Delayed Start (5min)
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 3
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 3
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 3
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 3
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 2
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 2
Apr 22 09:50:03 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 2
Apr 22 09:50:04 lldpd[289]: custom TLV op replace oui f8:32:e4 subtype 2
Apr 22 09:50:04 kernel: Init chrdev /dev/detector with major 190
Apr 22 09:50:04 kernel: tdts: tcp_conn_max = 8000
Apr 22 09:50:04 kernel: tdts: tcp_conn_timeout = 300 sec
Apr 22 09:50:20 kernel: SHN Release Version: 2.0.1 890c91d
Apr 22 09:50:20 kernel: UDB Core Version: 0.2.18
Apr 22 09:50:20 kernel: Init chrdev /dev/idpfw with major 191
Apr 22 09:50:21 kernel: IDPfw: IDPfw is ready
Apr 22 09:50:21 kernel: sizeof forward pkt param = 192
Apr 22 09:50:21 BWDPI: fun bitmap = ff
Apr 22 09:50:23 A.QoS: qos_count=0, qos_check=0
Apr 22 09:50:23 A.QoS: set_qos_on fails
Apr 22 09:50:23 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
Apr 22 09:50:23 kernel: ERR[qos_start:3364] qos_ops is not registered!
Apr 22 09:50:23 kernel: ioctl_iqos_op_switch(1) fail!
Apr 22 09:50:25 A.QoS: qos_count=0, qos_check=1
Apr 22 09:50:29 A.QoS: qos rule is less than 22
Apr 22 09:50:29 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
Apr 22 09:50:29 WLCEVENTD: eth1: Assoc
Apr 22 09:50:30 A.QoS: qos_count=1, qos_check=1
Apr 22 09:50:31 crond[239]: time disparity of 507105 minutes detected
Apr 22 09:50:34 dnsmasq-dhcp[231]: DHCPDISCOVER(br0)
Apr 22 09:50:34 dnsmasq-dhcp[231]: DHCPOFFER(br0)
Apr 22 09:50:34 dnsmasq-dhcp[231]: DHCPREQUEST(br0)
Apr 22 09:50:34 dnsmasq-dhcp[231]: DHCPACK(br0)
Apr 22 09:50:34 A.QoS: qos rule is less than 22
Apr 22 09:50:34 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
Apr 22 09:50:36 A.QoS: qos_count=2, qos_check=1
Apr 22 09:50:40 A.QoS: qos rule is less than 22
Apr 22 09:50:40 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
Apr 22 09:50:40 rc_service: udhcpc 398:notify_rc start_firewall
Apr 22 09:50:41 dhcp_client: bound  during 604800 seconds.
Apr 22 09:50:41 miniupnpd[566]: shutting down MiniUPnPd
Apr 22 09:50:41 nat: apply nat rules (/tmp/nat_rules_eth0_eth0)
Apr 22 09:50:42 custom_script: Running /jffs/scripts/firewall-start (args: eth0)
Apr 22 09:50:42 miniupnpd[2673]: HTTP listening on port 34237
Apr 22 09:50:42 miniupnpd[2673]: Listening for NAT-PMP/PCP traffic on port 5351
Apr 22 09:50:44 adaptive QOS: Delayed Start Canceled
Apr 22 09:50:46 adaptive QOS: Applying - Iptable Down Rules
Apr 22 09:50:47 adaptive QOS: Applying - Iptable Up   Rules (eth0)
Apr 22 09:50:47 adaptive QOS: TC Modification Delayed Start (5min)
Apr 22 09:50:57 WLCEVENTD: eth2: Assoc
Apr 22 09:50:58 dnsmasq-dhcp[231]: DHCPREQUEST(br0)
Apr 22 09:50:58 dnsmasq-dhcp[231]: DHCPACK(br0)
Apr 22 09:51:51 kernel: htb: htb qdisc 10: is non-work-conserving?
Apr 22 09:51:55 kernel: htb: htb qdisc 10: is non-work-conserving?
Apr 22 09:55:52 adaptive QOS: Applying  TC Down Rules
Apr 22 09:55:52 adaptive QOS: Applying  TC Up   Rules
Apr 22 09:55:52 adaptive QOS: Modifying TC Class Rates
Apr 22 09:55:52 kernel: HTB: quantum of class 10011 is big. Consider r2q change.
Apr 22 09:55:52 kernel: HTB: quantum of class 10015 is big. Consider r2q change.

 

[randomName]

Most smart devices won't be truly turned off unless you physically unplug them from their power outlet ,they just go into a low power state. My Mi box is always sending/receiving data unless I pull the plug.

We're living Orwellian times I'm afraid.

Your ISP was more than likely the cause of your drastic bufferbloat change.

Thanks for the reply. I was thrown for a loop when my main rig got an A+ then my Game rig immediately got a C. I don't know. Seems like constantly maintenancing the router seems to help, too. Lately the GUI seems to be freezing up requiring a hard reboot.

@FreshJR You've recommended keeping devices at default priority to keep a sane allotment of bandwidth for everyone, but in my case I dictate what devices get the amount of bandwidth given and that brings up some curiosity about how the Device Priority function works. Are simple changes just made or is there more going on than just that like CPU resource allocation, etc?

Thanks

 

[FreshJR]

@Sptz after the 5 minute wait, you should stop getting QOS related messages in the system log except for harmless quantum/non-work conserving messages.

Also in the debug mode you should look for undf flowid not being equal to 1:17, instead of undf prio (typo in first post i haven't updated)

If it keeps turning off and on, study rc_start messages to see what using action keeps triggering QoS. Qos works best after it settles down so it’s best if it’s not disturbed continuously.

@randomName I don’t see how you are dictating how much bandwidth devices get. Are you cascading two forms of QOS?

Device priority is achieved by another set of HTBs within each category of the main HTB.

Eg you have your main HTB with your 8 QoS categories.

Then within each category you have another HTB that each contians up to 255 devices. It’s good to keep all devices same priority since the rate/ciel selection is in the cascaded HTB is atrocious. (Just like the poor values in the main HTB before I modified them)

Simply put, a cascaded HTB is not the proper tool for the intended action of device priority. It should of have been a queue with different traffic processing logic (modified round robin).

 

[randomName]

@FreshJR Yes, I tried out the Device Priority by sliding 'Highest' (red) to my gaming rig. Just for curiosity does Apps Analysis have to be running for for a differently allocated device priority?

What's an HTB?

 

[FreshJR]

@FreshJR Yes, I tried out the Device Priority by sliding 'Highest' (red) to my gaming rig. Just for curiosity does Apps Analysis have to be running for for a differently allocated device priority?

What's an HTB?

You were curious how device priority works so I explained the current implementation.

No AppAnalysis does not have to be running.

No don't use it, it's doesn't work well no matter how hard you wish it works properly.
Feel free continuing to experiment so you can reach your own conclusion.

 

[Grisu]

@FreshJR What's an HTB?

Possibly this is meant with HTB: https://sourceforge.net/projects/htbinit/

 

[randomName]

Thanks @Grisu !

So I had to do a Restore to Default settings including initialize due to my WebGUI randomly freezing. Is 8.8 the latest?

 

[Vexira]

Thanks @Grisu !

So I had to do a Restore to Default settings including initialize due to my WebGUI randomly freezing. Is 8.8 the latest?

What ever is on the front page is the latest.

 

[Andy1932]

I just realized that on the Adaptive QOS/Bandwidth Monitor page, you can drag and drop priorities onto individual devices.
Should I not mess with this? Will it mess up the JR Fresh settings?

 

[Sinner]

I just realized that on the Adaptive QOS/Bandwidth Monitor page, you can drag and drop priorities onto individual devices.
Should I not mess with this? Will it mess up the JR Fresh settings?

Correct dont mess with those. Theyre not implemented properly and are unness as freshjr qos does the job sorta already by data type.

 

[JohnSmith]

Correct dont mess with those. Theyre not implemented properly and are unness as freshjr qos does the job sorta already by data type.

I have been setting those priorities on my devices on my network (mostly to band them together to visually see if a device is using a lot of bandwidth, and also pushing the devices I want to Highest Priority to see them at top of the list), and so far I have seen no difference in my network that this is causing an issue, but maybe Merlin or FreshJr could chime in to let us know if there is any issue tagging these priorities onto devices, as well as using FreshJr Adaptive QOS.

 

[randomName]

@JohnSmith post 3127 & 3128, I asked the same question and Fresh replied.

 

[Sinner]

If you read thru this thread theres approx 1 post from freah stating not to use it per page roughly.

 

[FreshJR]

so far I have seen no difference in my network that this is causing an issue

I have explained it many times. LACK OF SANE BANDWIDTH ALLOTMENT.

To see for yourself, simply run two speed tests simultanously on devices of different priorities.

- One device will get 99.9% of the bandwidth
- The other device will get 00.1% of the bandwidth.

If you think that is acceptable behavior and a functioning QoS system, then more power to you.

Now for devices of the same priority, the bandwidth allocation is alloted fairly.

 

[randomName]

Couple of questions.

When there is an update to FreshJR, how will we know?

Also, the past couple of days my Bufferbloat has been really good, 1 A, and the rest A+, but the odd thing is the idle, down and up, latency MS reduced by half. I normally have been getting around 28-30ms on both up and down, and the past 2 days there were in the mid teens (16-19ms) respectively while the idle ms was just tipping over 10ms. Is that normal or is something not reading correctly cause as of tonight the latency has gone back to what I've normally been seeing 28-30ms