[Beta] Asuswrt-Merlin 380.67 Beta is now available

[RMerlin]

I've just installed beta3b to my 88U. When I go to page QoS Statistics I think httpd crashes because I can't access to any page on web UI till restart httpd service.

and now cpu is %100 and for a while and here is htop output which seems weird to me.

https://prnt.sc/ft2r02

I've chosen "WAN packet overhead" as cable(DOCSIS) which I use. but after apllying this it still shows "Select Preset:"

CPU is %100 for about 10 minutes maybe more and router uptime is 13 minutes already. I will reboot it now.

Your router is getting stuck trying to run a bunch of "tc class change" calls. Sounds like the fq_codel patch is getting stuck in an endless loop. Are these from a custom script?

 

[Adamm]

Next update on ipset 6....Just couldn't leave it alone. Was able to isolate the crash to flushing settype hash only. Dug through the ipset code. Good news this time! After a bit of trial and error I came up with a patch for ipset that appears to fix the crash. Back on track for now....

In addition to comment (which I haven't tried yet, maybe you could help out? If need be, I can port it over to the Merlin Beta)....I think counters looks interesting...

admin@RT-AC68P-EC58:/tmp/home/root# ipset list | grep -v "packets 0 bytes 0"
Name: NoVPNBase
Type: hash:ip
Revision: 4
Header: family inet hashsize 4096 maxelem 65536 counters
Size in memory: 8100
References: 1
Number of entries: 67
Members:
52.94.236.178 packets 20 bytes 5463
52.43.24.139 packets 33 bytes 49629
176.32.103.58 packets 52 bytes 12818
52.24.246.177 packets 9 bytes 4217
54.239.17.71 packets 11 bytes 2286
52.94.236.223 packets 10 bytes 1269
176.32.103.194 packets 17 bytes 3680
54.239.27.11 packets 23 bytes 5877
54.239.16.1 packets 14 bytes 3560

Name: NoVPNCIDR
Type: hash:net
Revision: 6
Header: family inet hashsize 4096 maxelem 65536 counters
Size in memory: 1156
References: 1
Number of entries: 7
Members:
23.246.0.0/18 packets 41278 bytes 2809906   << Streaming Netflix

Name: NoVPN
Type: list:set
Revision: 3
Header: size 8 counters
Size in memory: 124
References: 1
Number of entries: 2
Members:
NoVPNBase packets 765 bytes 391807
NoVPNCIDR packets 41278 bytes 2809906

More then happy to test it out for you and give it a few real world tests. Preferably a version forked on merlins codebase would be ideal (so I don't need to reconfigure etc), but if its too much work I'd be fine jumping over to your fork to give it a whirl. Let me know

Thanks again for looking into it, already have a few ideas in mind how I can put it all to use.

 

[pattiri]

Your router is getting stuck trying to run a bunch of "tc class change" calls. Sounds like the fq_codel patch is getting stuck in an endless loop. Are these from a custom script?

yes there are custom scripts that are causing this but after a reboot it resolved. Thank you for your interest

 

[RMerlin]

yes there are custom scripts that are causing this but after a reboot it resolved. Thank you for your interest

I'll make some more changes to the patching process to make it more robust and avoid any potential loop like you experienced. I suspect your script was run at the wrong time, causing the patch to be re-applied on itself, and leading to these loops.

 

[RMerlin]

More then happy to test it out for you and give it a few real world tests. Preferably a version forked on merlins codebase would be ideal (so I don't need to reconfigure etc), but if its too much work I'd be fine jumping over to your fork to give it a whirl. Let me know

Thanks again for looking into it, already have a few ideas in mind how I can put it all to use.

I'll try to get these changes merged into Github during the weekend, so you'll be able to build your own image with it.

 

[Eric Lieb]

OT, but if you mean you have one server with redirect, and one without, you should consider doing this on the client side. Don't redirect on the server side, and set up two client configs, one for Starbucks with redirect, one for a secure location without. The other server is then available for something else. Like one TAP, one TUN, or one 443/TCP and one more flexible.

I am not sure that is a client side option... I have only seen references to it being an openvpn server setting.

https://openvpn.net/index.php/open-source/documentation/howto.html#redirect

that states you have to configure the server for this option since the server has to handle all the dns queries.

 

[el pescador]

Ive flashed the beta 3b.
Seen the wan packet overhead change.
If i have the isp router behind the asus which has a set wan address from dchp off the isp router....what setting do i use?

The isp router uses a hidden pppoe sign in and i think it uses vlans....has the phone connected to the isp router...

 

[Shonk]

Are you using a public TLD (.org) within your LAN? If so, that's a bad idea for many different reasons.

When I access my own router using its internal hostname I have no problem repeating the same steps as you.

yes there's nothing wrong with using a rfc 1918 ip with a public tld for my own internal use clearly the domain isnt only used for that
i have been doing this since around 2004 its fine

some code merges in 2014 broke it briefly and was fixed by yourself or asus shortly after
its again broken since last week
just pointing out there's a problem with the code somewhere is all

@3b3bf31 is the first build i did that had the problem on the 29th

if you are using the 192.168.0.x subnet try accessing it with aurora.shonk.org (resolves to 192.168.0.1)

 

[Swistheater]

Ok. I have been experiencing issues with the 380.67 beta 3b that seem to stem from the same problems of regular Beta 3. Once I have Selected adaptive and manual and have set my bandwidth, I am not able to save priorities under custom option. I have IE 11. It will not continue with the Save. I have the RT-3100. I have also been experiencing a slue of issues on IE 11 including browser crashes since the update to Beta 3. I am not even able to access this forum with out the browser crashing. I was not experiencing these issue on beta 2.

 

[RMerlin]

I am not even able to access this forum with out the browser crashing.

I suggest you start by debugging your browser, or switch to a different one. If even regular websites cause your browser to crash, then there's a problem with your browser or an installed add-in. The router has nothing to do with browser issues.

 

[Swistheater]

The only time i am having issues with browser crashes is when i try to load into this forum. no clue what is causing it. back to the router issue though. I am not able to save changes when modifying the custom preferences of the adaptive QoS unless i use firefox or chrome. I do not usually use these browsers to change any settings in the webui though. So I know this issue is router related. I am able to modify any other settings perfectly fine though.

https://ibb.co/k6jDKa

this is browser crash

https://ibb.co/fT7Hea

 

[elorimer]

I am not sure that is a client side option... I have only seen references to it being an openvpn server setting.

https://openvpn.net/index.php/open-source/documentation/howto.html#redirect

that states you have to configure the server for this option since the server has to handle all the dns queries.

Sorry about the OT. But if you don't push redirect_gateway from the server then doesn't it depend on whether the client has --redirect-gateway? I think I'm doing this with a Chromebook, where I have one .onc with "ignore_default_route" and one without. I think that is the equivalent for .ovpn.

 

[RMerlin]

The only time i am having issues with browser crashes is when i try to load into this forum. no clue what is causing it. back to the router issue though. I am not able to save changes when modifying the custom preferences of the adaptive QoS unless i use firefox or chrome. I do not usually use these browsers to change any settings in the webui though. So I know this issue is router related. I am able to modify any other settings perfectly fine though.

https://ibb.co/k6jDKa

this is browser crash

https://ibb.co/fT7Hea

IE is probably tripping on the broken DOM used by most of Asuswrt's webui. I got a workaround for the EZ-QoS page that will at least fix the Javascript error occuring when attempting to save priorities.

 

[RMerlin]

Ive flashed the beta 3b.
Seen the wan packet overhead change.
If i have the isp router behind the asus which has a set wan address from dchp off the isp router....what setting do i use?

The isp router uses a hidden pppoe sign in and i think it uses vlans....has the phone connected to the isp router...

I'd use the appropriate VDSL/ADSL profile, and possibly add 4 more bytes to take the VLAN into account. That's what I think however, can't say I'm really an expert in this area.

 

[cloneman]

Hi Merlin,

Just trying to wrap my head around QoS differences between versions/architectures. I'm usually a Tomato user, I'm following (from afar ) the progression of the QoS systems.


1. Has fq_codel been added to MIPS? I thought it was for ARM only... are their seperate changelogs for ARM/MIPS?
(I'm satisfied with SFQ, but that's because my VDSL modem has very little bufferbloat)

2. Is there a plan to implement imq/ifb inbound borrowing? When I last used Merlin it didn't have bandwidth borrowing between classes in inbound ; just hard ceilings, 1 single column in webui, as opposed to tomato shibby's Min/Max a.k.a. guaranteed/ceiling a.k.a tc rate/ceil

3. If no inbound inter-class borrowing, I assume the global download limit works ? (this would be required for fq_codel or even sfq to work on the downstream side of things, so it knows when to start dropping packets). In older Tomato QoS versions the global download limit was not a thing, it was just a a placeholder to calculate percentage ceilings a.k.a. policers for each class.

 

[RMerlin]

Has fq_codel been added to MIPS?

No, their kernel is too old.

are their seperate changelogs for ARM/MIPS?

No, same changelog for all models. Unlike Tomato, I don't have separate codebases for MIPS and ARM.

Is there a plan to implement imq/ifb inbound borrowing?

Not possible. IMQ requires kernel structural changes that are not compatible with the closed-source components.

 

[lilstone87]

Merlin,

I noticed something odd, and likely nothing you can do. But i wanted to mention it. With Adaptive QoS enabled, with fd_codel selected, and bandwidth limits manually set. When i run the speedtest on dslreports, the test oddly bypasses the limits i inputted. But if i go to other speedtest sites, when running the test, speed's stay below the limits i set correctly.

I know with the recent changes you have made with QoS, there has been others testing, and even posting speedtest results from dslreports. So i wonder when others run this test, is it honoring the limits you set within QoS on the router?

 

[Vexira]

Ive flashed the beta 3b.
Seen the wan packet overhead change.
If i have the isp router behind the asus which has a set wan address from dchp off the isp router....what setting do i use?

The isp router uses a hidden pppoe sign in and i think it uses vlans....has the phone connected to the isp router...

vdsl 2 or fiber, if its vdsl2 use vdsl ipoe, isp dhcp is ipoe.
that's a preset idea for merlin ipoe over optic fiber, cause I know that's what my isp uses and if I get an optic fiber to the home upgrade or move to an area that has optic fiber and I'm on the same isp, the use ipoe.

 

[stylish_me]

Anyone has dual wan working? Mine stopped working after recent updates. Both connections are wired via lan cable, failover and failback are turned on. As well as watchdog and ping to internet.
As soon as i unplug primary wan cable - router switches to secondary wan. But when primary wan cable is plugged in and my provider goes down - router doesnt switch to secondary wan. It even doesnt realise that it cannot ping 8.8.8.8 - router just stays offlne.
Would like to make full reset to be sure it is firmware related problem. But i have over 40 manually assigned ip's and names to my network devices - and i am not really sure how to backup this information.
So if anyone has dual wan with ping to internet and both wired connections fully working - i will have to make factory reset to fix it. But if it is not working for everyone - i will not bother. Thanx!

 

[Adamm]

I'll try to get these changes merged into Github during the weekend, so you'll be able to build your own image with it.

What OS do you use to build with? I shut down my build server around a year ago, and trying to compile from source on any fresh ubuntu image (12.04 > 16.04) seems to be a nightmare of endless errors