[Release] Asuswrt-Merlin 384.11 is available

[RMerlin]

@RMerlin sorry to bug ya bro. Was curious of your fq_codel implemtation built in. What is the target timing you have set or is it dynamic at all?

It's left to the kernel default values (i.e. it's not specified at rule creation time).

Also i found once but cant seem to now. Whats the grep command for viewing which is being used on each interface? Like sfq fifo fq_codel etc? There was a command that showed all interfaces which would be handy for me atm while im diggin

tc qdisc

 

[Gar]

the only use you get out of it is redundancy. there are certain features that one dns server may have that another does not, that don't compliment each other.

for example filtered v.s. non-filtered, logged v.s. unlogged, qname-minimization-vs no qname-minimization.

the question remains are you losing something with one only to gain a little redundancy.

Does it stay with resolver 1 until there is a problem? What makes it look at resolver 2? Or am I misunderstanding what this process does. When you said compliment each other I got lost.

 

[Swistheater]

Does it stay with resolver 1 until there is a problem? What makes it look at resolver 2? Or am I misunderstanding what this process does. When you said compliment each other I got lost.

Round robin shares the query load
Round robin in stubby =1,
no round robin means only one sever gets used until it fails round robin in stubby =0

 

[Sinner]

It's left to the kernel default values (i.e. it's not specified at rule creation time).

tc qdisc

ok perfect I do see its using the standard default of 5secs. is there a way I could increase that:

qdisc fq_codel 1256: dev eth0 parent 10:256 limit 10240p flows 1024 quantum 1514 target 5.0ms interval 100.0ms

I have a particularly slower cellular connection sub 10mb and with some research im finding out that it takes longer that 5ms to communicate and id like to mess with increasing this slightly if possible. I realize the qdiscs are created on the fly for each connection so im wonderinf if there was like a primary setting or change I could make or add to freshjr's script temporarily so default or newly created qdiscs would have increased target time on the fq_codels?

 

[Gar]

Round robin shares the query load
Round robin in stubby =1,
no round robin means only one sever gets used until it fails round robin in stubby =0

So using DoT in Merlin f/w it is set to use Rr by default?

 

[Swistheater]

So using DoT in Merlin f/w it is set to use Rr by default?

Yes

 

[Sinner]

im jus shooting in the dark here trying to find a way to get any modifications to stick but nothing working so far! I feel like im close with this lol:

#tc qdisc add dev eth0 root fq_codel limit 10240 target 40ms interval 120ms ecn
#tc qdisc add dev br0 root fq_codel limit 10240 target 40ms interval 120ms ecn

anyone have any suggestions?

*edit*
think I got it.. had to remove the pound symbols lol and remove the previous roots. testing now

ok it appears to be working but messes up the qos statistics page. it may be deleting all the rules this is a little above me but testing anyway

 

[RMerlin]

I uploaded test builds to https://asuswrt.lostrealm.ca/test-builds . Please confirm that nothing is horribly broken in them.

  - NEW: Implemented source/destination IP filtering
         for the Netool version of Netstat web page.
  - CHANGED: Backported multiple fixes and improvements
             for ntpd from upstream, improving handling
             of failed server hostname resolution, and better
             clock sync discipline.
  - FIXED: RT-AC88U/3100/5300 were accidentally compiled
           with Netool enabled, which isn't compatible with
           these model's kernel.
 - FIXED: Movistar stopped working for some users.  Re-disabled
          udpxy on Movistar profile for now.  A more complete
          fix will have to come from Asus.
 - FIXED: Re-disabled memaccess debugging tool, as it creates
          a symlink called "sh" which is a pretty bad
          idea from Broadcom. (RT-AC86U, RT-AX88U)

 

[skeal]

Just flashed my AX88U. It took forever for the ntp to update. I had a DHCP IP and the connection showed "up" and still no time sync. I loaded the log page and waited, and waited, and then finally the ntp update happened and everything works ok apart from that.

 

[Butterfly Bones]

FIXED: Re-disabled memaccess debugging tool, as it creates a symlink called "sh" which is a pretty bad idea from Broadcom. (RT-AC86U, RT-AX88U)

Thank you, this should help us RT-AC86U script users.

:/tmp/home/root# memaccess /jffs/scripts/vpnclient1-route-up
-sh: memaccess: not found

 

[Kingp1n]

I uploaded test builds to https://asuswrt.lostrealm.ca/test-builds . Please confirm that nothing is horribly broken in them.

  - NEW: Implemented source/destination IP filtering
         for the Netool version of Netstat web page.
  - CHANGED: Backported multiple fixes and improvements
             for ntpd from upstream, improving handling
             of failed server hostname resolution, and better
             clock sync discipline.
  - FIXED: RT-AC88U/3100/5300 were accidentally compiled
           with Netool enabled, which isn't compatible with
           these model's kernel.
 - FIXED: Movistar stopped working for some users.  Re-disabled
          udpxy on Movistar profile for now.  A more complete
          fix will have to come from Asus.
 - FIXED: Re-disabled memaccess debugging tool, as it creates
          a symlink called "sh" which is a pretty bad
          idea from Broadcom. (RT-AC86U, RT-AX88U)

RMerlin....sorry to bug you...I know you mentioned you're waiting on the new GPL for the latest asus FW released... is this something you think we should be worrying about in the near future? Or once the GPL is realeased we can have a better picture on how it may affect all of us? Basically should we be concerned?

 

[scjr]

Flashed to 384.11_1. All good thus far.

 

[Swistheater]

I uploaded test builds to https://asuswrt.lostrealm.ca/test-builds . Please confirm that nothing is horribly broken in them.

  - NEW: Implemented source/destination IP filtering
         for the Netool version of Netstat web page.
  - CHANGED: Backported multiple fixes and improvements
             for ntpd from upstream, improving handling
             of failed server hostname resolution, and better
             clock sync discipline.
  - FIXED: RT-AC88U/3100/5300 were accidentally compiled
           with Netool enabled, which isn't compatible with
           these model's kernel.
 - FIXED: Movistar stopped working for some users.  Re-disabled
          udpxy on Movistar profile for now.  A more complete
          fix will have to come from Asus.
 - FIXED: Re-disabled memaccess debugging tool, as it creates
          a symlink called "sh" which is a pretty bad
          idea from Broadcom. (RT-AC86U, RT-AX88U)

I tried it out for a few days now with self compile/merge, the only issue I saw was the ability to turn on and off updates checks has been removed.

As for some of the fixes I.e. Traceroute is normal now and the ntp server seems to be functioning as it should.

I will flash these new compiles later and see how it goes.

 

[Swistheater]

Just flashed my AX88U. It took forever for the ntp to update. I had a DHCP IP and the connection showed "up" and still no time sync. I loaded the log page and waited, and waited, and then finally the ntp update happened and everything works ok apart from that.

I noticed this with the first reboot after update, but afterwards everything ran smoothly with following reboots.

 

[Swistheater]

I can't wait to test 384.12 once it is viable and ready. Merlin keeps rolling out the next best thing.

 

[Sinner]

ok perfect I do see its using the standard default of 5secs. is there a way I could increase that:

qdisc fq_codel 1256: dev eth0 parent 10:256 limit 10240p flows 1024 quantum 1514 target 5.0ms interval 100.0ms

I have a particularly slower cellular connection sub 10mb and with some research im finding out that it takes longer that 5ms to communicate and id like to mess with increasing this slightly if possible. I realize the qdiscs are created on the fly for each connection so im wonderinf if there was like a primary setting or change I could make or add to freshjr's script temporarily so default or newly created qdiscs would have increased target time on the fq_codels?

im probably not doing it correctly but it seems as thought this mod cant be applied from freshjr qos script at least. possibly before it runs someplace else?
I realize this isn't the qos thread but this is more of a merlin question. id just like it applied for qos.

 

[Gitsum]

I uploaded test builds to https://asuswrt.lostrealm.ca/test-builds . Please confirm that nothing is horribly broken in them.

  - NEW: Implemented source/destination IP filtering
         for the Netool version of Netstat web page.
  - CHANGED: Backported multiple fixes and improvements
             for ntpd from upstream, improving handling
             of failed server hostname resolution, and better
             clock sync discipline.
  - FIXED: RT-AC88U/3100/5300 were accidentally compiled
           with Netool enabled, which isn't compatible with
           these model's kernel.
 - FIXED: Movistar stopped working for some users.  Re-disabled
          udpxy on Movistar profile for now.  A more complete
          fix will have to come from Asus.
 - FIXED: Re-disabled memaccess debugging tool, as it creates
          a symlink called "sh" which is a pretty bad
          idea from Broadcom. (RT-AC86U, RT-AX88U)

What is different between 384.11_1 and 384.12 alpha1 for the AX88U ?

 

[rk8531]

Installed the test build and everything is working back to normal. Logs are much more cleaner than before.

 

[galen2]

Movistar IPTV its working again on 348.11_1 test build

 

[Odkrys]

@RMerlin
wsdd2 is super useful. thanks for implement.