What's new

Beta Asuswrt-Merlin 386.5 Beta is now available

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Status
Not open for further replies.
Hmm, maybe this rev does not like having more than 1 dns target to verify whether it is the MS dns or not? As soon as I added a 2nd dns target with associated IPs, the main GUI reverted to "Disconnected".
Outcome: select 1 dns target and it's IP. I had no issues still listing multiple ping targets.
I don't know what "DNS target" you're referring to? If you mean the "Resolve Hostname" field that's not a "target" and it's meant to be a single host. There should really be some extra validation to stop that field accepting more than one string.
 
OK - so no other testers ... crash test dummy to the rescue ;) ... applies to RT-AX86U can't comment on other routers ...
  1. Traditional QoS works as never before - scores A+ on https://www.waveform.com/tools/bufferbloat? and has low latency. Note however that like Cake - it disables HW acceleration - both Runner and Flow Cache!
  2. Cake QoS works as well as it ever did.
  3. Adaptive QoS is definitely broken ... [which means FlexQoS which depends on Adaptive QoS - is also broken @dave14305 ].
Adaptive QoS -
  • Keeps HW acceleration enabled - Runner and Flow cache;
  • "Bandwidth Monitor" does not provide accurate speed readings;
  • Bufferbloat scores a "C" if you lucky;
  • Latency scores are spread all over the place;
Log shows the following ...
Code:
RT-AX86U-E330 A.QoS: qos_count=0, qos_check=0
RT-AX86U-E330 A.QoS: qos rule is less than 22
RT-AX86U-E330 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
RT-AX86U-E330 A.QoS: qos_count=0, qos_check=1
RT-AX86U-E330 A.QoS: qos rule is less than 22
RT-AX86U-E330 A.QoS: restart A.QoS because set_qos_conf / set_qos_on / setup rule fail
RT-AX86U-E330 kernel: [tdts_shell_ioctl_stat:256] Recv ioctl req with op 2

Adaptive QoS worked really well under stock 46061 and under Merlin 386.4
This is of concern for me, Flex is pretty much my number 1 reason for having Merlin, is Flex broken in this release?

If so I will continue to avoid as on 386.3_2 everything is working well, uptime 40 days +
 
I don't know what "DNS target" you're referring to? If you mean the "Resolve Hostname" field that's not a "target" and it's meant to be a single host. There should really be some extra validation to stop that field accepting more than one string.
Hi - Sorry. My bad. I should have been clearer... "DNS target" is "Resolve Hostname" ... I had a list of "Resolve Hostname" separated by spaces and their matching "Resolved IP Addresses"... Maybe it only ever used the first one, IDK for sure.. Now using only 1 seems to be the requirement. See the example "Ping Target"... which still seems happy with. multiples.
1645894628158.png


Update: After the discussion here, I removed all but 1 target from the fields above. At some point, I can swear I read somewhere or someone posting something said that you could list multiples, separated by spaces in case that single test target went down for some reason. TY for the clarifications!
 
Last edited:
Dang! it has been four days on this firmware with no changes or reboot. Guess it is time to open the job jar for a home project to work on. Maybe I can try WIndows 11 on my desktop again? Nah, am getting sick of Microsoft and their approved hardware. Merlin, however, keeps getting better!
 
Hi - Sorry. My bad. I should have been clearer... "DNS target" is "Resolve Hostname" ... I had a list of "Resolve Hostname" separated by spaces and their matching "Resolved IP Addresses"... Maybe it only ever used the first one, IDK for sure.. Now using only 1 seems to be the requirement. See the example "Ping Target"... which still seems happy with multiples. View attachment 39832
I suspect it was only ever using one host name, or at least I think that was the intention. It kinda doesn't make sense with what it's meant to be doing otherwise.

Anyway, I believe it is this patch that "fixed" the old behaviour which was presumably treating the contents of that field as one single hostname that contained spaces (which of course is not allowed).

Specifying more than one hostname or IP address in the Ping Target field also results in a false "network disconnected" message for me, unless the DNS Query is enabled and successful. Again, looking at wanduck.c it appears that this field is only intended to have one hostname or IP address. Having more than one would create a syntactically invalid ping command.

@RMerlin Can you confirm that the "Resolve Hostname" and "Ping Target" fields are only meant to contain a single entry. If so, perhaps they should not accept input that contains spaces. Same for the Dual WAN page.
 
This is of concern for me, Flex is pretty much my number 1 reason for having Merlin, is Flex broken in this release?

If so I will continue to avoid as on 386.3_2 everything is working well, uptime 40 days +

Agreed FlexQOS is the main draw for me, along with DNSFilter. It would be disappointing if Adaptive QOS is broken for whatever reason. I wish I could help test but the protests that come along with a reset and setup are too much at the moment.
 
Something looks wrong with bwdpi and flow cache on the 5.02p1 platform (RT-AX68U and RT-AX86U). I can try to poke at Asus about it next week, but otherwise not much I can do about this since both are outside of my control. Same issue also exists on the stock firmware.
 
Something looks wrong with bwdpi and flow cache on the 5.02p1 platform (RT-AX68U and RT-AX86U). I can try to poke at Asus about it next week, but otherwise not much I can do about this since both are outside of my control. Same issue also exists on the stock firmware.
Are there any negative side effects to be expected outside of Adaptive QoS since you mentioned flow cache?
 
Dang! it has been four days on this firmware with no changes or reboot. Guess it is time to open the job jar for a home project to work on. Maybe I can try WIndows 11 on my desktop again? Nah, am getting sick of Microsoft and their approved hardware. Merlin, however, keeps getting better!
Make a dual boot linux windows machine, that could be fun :)
 
Today I saw this in syslog:
Code:
Feb 26 12:14:31 kernel: bcm_mcast_mld_add:833 mc_fdb->rep_list ffffffc013532728 next ffffffc007397420 prev ffffffc007397420 rep_entry->list ffffffc007397420 next ffffffc013532728 prev ffffffc013532728
What's this?
Look here:


I am still trying to figure out what's causing this, but apart from some useless conversation in this thread i still dont know.
 
Are there any negative side effects to be expected outside of Adaptive QoS since you mentioned flow cache?
Probably affects everything tied to the Trend Micro BWDPI engine. No idea if FC itself is working properly or not.
 
Dirty upgrade from 386.5 Alpha2 to Beta1. Three points of note...
1. I no longer unmount the USB flashdrive that has all the Merlin-Addon "stuff" (lazy and less time saving steps). The code of recent 386.4/386.5 seems to deal with it by itself. Previously (six months ago or later...) I usually couldn't add the firmware to the RT-AX88U without doing an unmount first.
2. After the update I forgot to reboot the router 30-60mins later in my method to recycle the unit in all aspects (power off/on later on an hour later). As like the previous two times with 386.4/.5 code I forgot to reboot/power off/on, within a day the WAN red light comes on as no Internet connection. After a reboot though, no issues with WAN connection.
3. I haven't had to time to roll-out the primary RT-AX88U router and install a GT-AXE11000 in it's place. I just have the AXE11000 (and spare) as AiMesh units to give them power on runtime, with the Beta1 code...but...they will drop from the Wireless connected Mesh within a few hours. Power off/on and they re-join. Done WPS button hold down on the units with power off, after code updates, and they still will drop from Mesh. None of the other units AX56U, AX88U, AX86U, will drop from the AiMesh. Really implies a ASUS Mesh code bug with the AXE11000?
All working great in a busy home & home/work environment. RMerlin plus his accomplices making this all occur for us, just awesome!
 
I suspect it was only ever using one host name, or at least I think that was the intention. It kinda doesn't make sense with what it's meant to be doing otherwise.

Anyway, I believe it is this patch that "fixed" the old behaviour which was presumably treating the contents of that field as one single hostname that contained spaces (which of course is not allowed).

Specifying more than one hostname or IP address in the Ping Target field also results in a false "network disconnected" message for me, unless the DNS Query is enabled and successful. Again, looking at wanduck.c it appears that this field is only intended to have one hostname or IP address. Having more than one would create a syntactically invalid ping command.

@RMerlin Can you confirm that the "Resolve Hostname" and "Ping Target" fields are only meant to contain a single entry. If so, perhaps they should not accept input that contains spaces. Same for the Dual WAN page.
You are awesome to keep trying to fight this fight - I appreciate it. I have given up and gone back to 386.3.2. Anything I do with Dual WAN and this "Disconnected" mess gives me nothing but frustration and headaches. Please let me know if you find a solution via PM. Ever since RT-AX88U_386.4_alpha2-ga85421da82

And yes, I updated to 386.5 beta 1 - back to internet disconnected.
Yes I've factory wiped
Yes I've done the nvram trick
NOTHING WORKS but going to the above alpha or earlier on Dual WAN Load Balance
 
Last edited by a moderator:
This is of concern for me, Flex is pretty much my number 1 reason for having Merlin, is Flex broken in this release?
FlexQoS definitely broken - but through no fault of FlexQoS ! Until Adaptive QoS [Flow Cache?] is fixed by Asus - best stay on prior version if FlexQoS essential for your system.

I have found that Cake works 100% for my needs - and also means I can completely avoid TrendMicro [not obsessed with Privacy issues - just don't see the need to burden my system].
 
So when using VPN and VPN director on a dual-wan loadbalancing system the VPN shows connected. And uses full bw of wan1 and wan2 but the VPN is connected to only 1 connection. So second connection is open wan. I did not notice this before is there a way to tell Merlin to use one off the two wans ore something
 
Adaptive QoS always broke for me, and fully reseting usually fixed it or manually setting the bandwith limits, not sure why i even bother leaving that on at this point, not a fan of Asus software team at all, they mess up everything all the time including armoury crate which pegs 50% cpu usage on 1 core for me when i sync my rgb for my memory, just to give you an idea how bad i consider Asus software team.
Adaptive QoS has been buggy since if owned my AX88U but does work usually after some reseting.
 
You are awesome to keep trying to fight this fight - I appreciate it. I have given up and gone back to 386.3.2. Anything I do with Dual WAN and this "Disconnected" mess gives me nothing but frustration and headaches. Please let me know if you find a solution via PM. Ever since RT-AX88U_386.4_alpha2-ga85421da82

And yes, I updated to 386.5 beta 1 - back to internet disconnected.
Yes I've factory wiped
Yes I've done the nvram trick
NOTHING WORKS but going to the above alpha or earlier on Dual WAN Load Balance
Sorry @scott420 but I think you need to read my post again. I'm not "fighting" anything. I was just pointing out that it is my understanding that the two fields mentioned above should only contain one entry otherwise they will create false errors.
 
FlexQoS definitely broken - but through no fault of FlexQoS ! Until Adaptive QoS [Flow Cache?] is fixed by Asus - best stay on prior version if FlexQoS essential for your system.

I have found that Cake works 100% for my needs - and also means I can completely avoid TrendMicro [not obsessed with Privacy issues - just don't see the need to burden my system].
mmm, thanks, not sure if Cake is good for 500/30. Was ok when I had 200/30. Hope they fix this asap. This is frustrating, all firmwares up to where I am updated fine until the last couple with old settings overwritten as well. Last one, I believe you have to reset and do all the settings manually, and this one flex is broken. Let's hope asus fix this.
 
Just one of the reasons I went with a dedicated firewall/router solution and use my old Asus for wifi duties only. :)
 
Status
Not open for further replies.

Support SNBForums w/ Amazon

If you'd like to support SNBForums, just use this link and buy anything on Amazon. Thanks!

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top