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

[FreshJR]

some of them have an * in DHCP table, some have proper names there.

I can get populate the names with the fuzzy values from dhcp/arp until clientlidt updates itself. “*” instead of device name included.

Can you check if arp from terminal is more no reliable with device names?

 

[JohnSmith]

Somethings wrong with my install on AX88U.

I can confirm Gitsum issue as well on my RT-AX88U as well, as I get the following issues on the QOS Classification Page:

1. My Classification Page looks like this from IE 11, and locks up the web page (refresh the IP address only, without the ASP page allows me to get back to the GUI):

Also, the web page link when you highlight says the following (including spelling mistakes):
"Plese reconfigure the local IP address for your RT-AX88U, and ensure that it is different from your WAN's gateway.EX: 192.168.X http://<RT-AX88U IP Address>/QOS_Stats.asp"

From Google Chrome, it looks like Gitsum's issue as well, and much like the same issue that I reported in the beta version:

 

[Milan]

arp is showing a lower amount of clients as DHCP table, but clients with * are in arp output with "?"

 

[consorts]

does lan qos only limit lan:wan traffic? or it is also supposed to impact lan:lan (wifi to lan based nas).

 

[Milan]

hm, another thing, in the client list is my computer identified by name of the creator of network card instead of assigned network name...

 

[JohnSmith]

It's bugged - the script is actually looking for these:
View attachment 16360
fix like so:

Still have this issue on the latest version FYI, where when you press "2" you get (3) "custom QOS rules", if you press "9", you get the "update" prompt

 

[JohnSmith]

Also, in your script, for the Gaming rule, you have Port 433, which should be Port 443 in your "iptables -A" code

##DOWNLOAD (INCOMMING TRAFFIC) CUSTOM RULES END HERE  -- legacy
       
        if [ "$( echo $gameCIDR | tr -cd '.' | wc -c )" -eq "3" ] ; then
            iptables -D POSTROUTING -t mangle -o br0 -d $gameCIDR -m mark --mark 0x80000000/0x8000ffff -p tcp -m multiport ! --sports 80,443  -j MARK --set-mark ${Gaming_mark_down} &> /dev/null        #Gaming - (Incoming "Unidentified" TCP traffic, for devices specified, not from WAN source ports 80 & 443 -->  Gaming)
            iptables -A POSTROUTING -t mangle -o br0 -d $gameCIDR -m mark --mark 0x80000000/0x8000ffff -p tcp -m multiport ! --sports 80,433  -j MARK --set-mark ${Gaming_mark_down}

            iptables -D POSTROUTING -t mangle -o br0 -d $gameCIDR -m mark --mark 0x80000000/0x8000ffff -p udp -m multiport ! --sports 80,443  -j MARK --set-mark ${Gaming_mark_down} &> /dev/null        #Gaming - (Incoming "Unidentified" UDP traffic, for devices specified, not from WAN source ports 80 & 443 -->  Gaming)
            iptables -A POSTROUTING -t mangle -o br0 -d $gameCIDR -m mark --mark 0x80000000/0x8000ffff -p udp -m multiport ! --sports 80,433  -j MARK --set-mark ${Gaming_mark_down}

 

[FreshJR]

@consorts it only limits WAN traffic. LAN2LAN is allowed not theottled.

@JohnSmith got it, thank for additional reports. Existing bugs will remain till next update.

 

[JohnSmith]

I can confirm Gitsum issue as well on my RT-AX88U as well, as I get the following issues on the QOS Classification Page:

1. My Classification Page looks like this from IE 11, and locks up the web page (refresh the IP address only, without the ASP page allows me to get back to the GUI):
View attachment 16388

Also, the web page link when you highlight says the following (including spelling mistakes):
"Plese reconfigure the local IP address for your RT-AX88U, and ensure that it is different from your WAN's gateway.EX: 192.168.X http://<RT-AX88U IP Address>/QOS_Stats.asp"

From Google Chrome, it looks like Gitsum's issue as well, and much like the same issue that I reported in the beta version:

View attachment 16392

View attachment 16393

Was able to fix the report page for Chrome only (IE 11 still hangs for me), by changing the following code in the QOS_Stats.asp page under /www directory

Old Code:

var category_title = ["Net Control Packets", "<#752#>", "<#762#>", "<#756#>", "<#764#>", "<#751#>", "<#757#>", "Game Transferring"];

New Code:

var category_title = ["Net Control Packets", "<#756#>", "<#766#>", "<#760#>", "<#768#>", "<#755#>", "<#761#>", "Game Transferring"];

See if that works for you FreshJR.

 

[FreshJR]

Was able to fix the report page for Chrome only (IE 11 still hangs for me), by changing the following code in the QOS_Stats.asp page under /www directory

Old Code:

var category_title = ["Net Control Packets", "<#752#>", "<#762#>", "<#756#>", "<#764#>", "<#751#>", "<#757#>", "Game Transferring"];

New Code:

var category_title = ["Net Control Packets", "<#756#>", "<#766#>", "<#760#>", "<#768#>", "<#755#>", "<#761#>", "Game Transferring"];

See if that works for you FreshJR.

John I understood what the bug was when it was first reported.
(Gitsum reported it earlier in this post).

The real issue is that different routers have different references to these QoS strings passed through via ASP Code Nuggets.

If I use your updated references it would be broken for the other half of users where it is current to working.

The solution is either to

1) have new webpage pull data from the old webpage

2) Hardcode the class names in English.

I already fixed the webpage. I chose to hardcode references since I didn’t want to translate newly created “GameTransferring” into many languages.)

I was just waiting back to hear on two more bugs before pushing the release.

1) device name (resolved this afternoon-> fall back to fuzzy match when device entry missing in clientlist)
2) need someone on stock firmware to check if “service restart-firewall;restart-qos” works or if it’s a RMerlin feature (pending -> I don’t want to downgrade to test this one short command )

There won’t be a long wait until the fixed release so don’t stress. (I don’t want to push many incremental releases per bug vs one update that fixes everything)

 

[Kingp1n]

John I understood what the bug was when it was first reported.
(Gitsum reported it earlier in this post).

The real issue is that different routers have different references to these QoS strings passed through via ASP Code Nuggets.

If I use your updated references it would be broken for the other half of users where it is current to working.

The solution is either to

1) have new webpage pull data from the old webpage

2) Hardcode the class names in English.

I already fixed the webpage. I chose to hardcode references since I didn’t want to translate newly created “GameTransferring” into many languages.)

I was just waiting back to hear on two more bugs before pushing the release.

1) device name (resolved this afternoon-> fall back to fuzzy match when device entry missing in clientlist)
2) need someone on stock firmware to check if “service restart-firewall;restart-qos” works or if it’s a RMerlin feature (pending -> I don’t want to downgrade to test this one short command )

There won’t be a long wait until the fixed release so don’t stress. (I don’t want to push many incremental releases per bug vs one update that fixes everything)

I can 99% confirm that the service restart-firewall;restart-qos works on stock before I turned my AC68P to a node but I'm like you and don't want to downgrade my node to test. Hope someone can report this with 100% assurance.

 

[Asad Ali]

Done on ASUS RT-AC86U with QoS enabled on FreshJR default settings. [emoji16][emoji16][emoji573]

 

[Ava Pug]

thanks for this new update. it makes adding new rules so much user friendly.

plex working perfectly

 

[Vexira]

It would be nice if I could set the reserved total bandwidth form the GUI now, it would help me with my experiments of changing bettween 80-95% of the total bandwidth.

I still haven't settled on which was better yet though in my case 90% looks good or responsive compared to 95% which I half suspect is realted to my 88u's older kernel.

 

[FreshJR]

It would be nice if I could set the reserved total bandwidth form the GUI now, it would help me with my experiments of changing bettween 80-95% of the total bandwidth.

I still haven't settled on which was better yet though in my case 90% looks good or responsive compared to 95% which I half suspect is realted to my 88u's older kernel.

I don’t want to implement a second slider that will take only take 85-95% of the initial input.

I think is more confusing than following the existing instructions.

 

[Vexira]

I don’t want to implement a second slider that will take only take 85-95% of the initial input.

I think is more confusing than following the existing instructions.

All good, still would have been nice for a box that a manual percentage could be put in I suppose it would be less confusing with a drop down box with values form 80-95% like the per packet overhead one for reserved bandwidth but I understand what you mean.

 

[chris.at]

I don’t want to implement a second slider that will take only take 85-95% of the initial input.

I think is more confusing than following the existing instructions.

Good decision FreshJR. Different technologies have different overhead and therefore need different percentages. Everyone should be able to take a calculator and do simple maths on its own...

 

[Vexira]

Good decision FreshJR. Different technologies have different overhead and therefore need different percentages. Everyone should be able to take a calculator and do simple maths on its own...

The overhead is fixed Merlin has an overhead value for each connection type, ddwrt had fixed percentaged for bandwidth reserved, it's actually more complicated to therefore ask people to do something that can be done In the script especially when you factor in obscure speed test values that don't easily round to whole numbers easily.

Not everything is simple maths, I've seen enough new users who just input the values without doing that calculations expecting it to work granted there is instructions I personally don't expect people to read them or else there would be a lot less request for help.

 

[L&LD]

.

I may have messed something up, I was playing with the download and upload percentages. Made sure that they add up to 100% though. But it shouldn't have any effect on the observation below, right?

My observation is that speedtest.net isn't being classified as a download or upload as noted in the first post. I am seeing it in the web surfing class for both up and down. Maybe a bug?

 

[L&LD]

The overhead is fixed Merlin has an overhead value for each connection type, ddwrt had fixed percentaged for bandwidth reserved, it's actually more complicated to therefore ask people to do something that can be done In the script especially when you factor in obscure speed test values that don't easily round to whole numbers easily.

What connection type would I use for Fibre? I don't' see it in the list.