[Official Release] AiMesh Firmware v3.0.0.4.384.10007 for All Supported Products

[mogulman]

I am having the same problem. This has been an issue for some time.

I had that issue when I first started up my RT-AC86u. I would see Obi202 3-4 times with different IP addresses and different MAC addresses. I think I manually renamed things, once I knew what they were.

 

[Ronald Schwerer]

I did a full factory reset (pressing the button) after installing the latest firmware on my 86U. Also did that after I had this problem and it still did not help. I did not try reseting the client list, figured a full reset and configure would do that (since my customizations were zapped by the reset). Only thing I have not tried was a reset and not doing anything to the client list. That might be an interesting experiment.

Do you have a large number of clients? Maybe the list gets full (size wise) immediately. I had about 22 clients and have some of the names kind-of long. I just went in to shorten the more verbose ones.
If you use the nvram commands I show above, the list will get rebuilt immediately after the reboot. The default name is sent from the client. If there are lots of client devices, maybe temporarily reduce the number until you can shorten the longer strings.

 

[mogulman]

@arthurlien Anything else I can do to help debug this? The feedback option in my router RT-AC86u doesn't seem to work. It tells me that it can't connect and to copy and paste text into an email.

This happened to me today again...

• PC1 already booted up
• PC2 press power button to resume from sleep
• RDP into PC2 into PC1.
• Remembered the issue above, so instead, closed RDP Session and worked on PC2 directly.
• Pressed the power button on PC2 to suspend/sleep.
• PC1's connection to the network stopped.
  
• Powered on PC2 and PC1 started working again.
• Later on, Powered off PC2 and PC1 was fine.

I know this could be related to my computers, but this never happens without AiMesh and never happened when I had just the RT-AC1900p alone. FYI..These are both wired GB Ethernet to the main router.

This is happening now without RDP even being involved. Log with notes. Seems like AiMesh (over ethernet) seems to reconnect whenever my PC reconnects. Here is the log with my notes while it happened. If I don't do anything, it seems like it takes 5 minutes for PC1 to reconnect on it's own. I sent you a copy of this with MAC addresses via the Feedback email.

*** Cleared Log, Resume from Sleep PC2 (30:e1:71:xx:xx:xx)
Jan 31 14:00:34 kernel: eth4 (Ext switch port: 3) (Logical Port: 11) Link UP 1000 mbps full duplex
*** Standby/Sleep PC2
Jan 31 14:01:00 kernel: eth4 (Ext switch port: 3) (Logical Port: 11) Link DOWN.
*** PC1 (ec:f4:bb:xx:xx:xx) loses network connection, pings to router timeout, but still shows link up from PC.
Jan 31 14:01:00 kernel: net_ratelimit: 134 callbacks suppressed
Jan 31 14:01:05 kernel: net_ratelimit: 181057 callbacks suppressed
Jan 31 14:01:10 kernel: net_ratelimit: 169700 callbacks suppressed
Jan 31 14:01:15 kernel: net_ratelimit: 176202 callbacks suppressed
Jan 31 14:01:20 kernel: net_ratelimit: 179251 callbacks suppressed
Jan 31 14:01:25 kernel: net_ratelimit: 180408 callbacks suppressed
Jan 31 14:01:30 kernel: net_ratelimit: 170621 callbacks suppressed
Jan 31 14:01:35 kernel: net_ratelimit: 175121 callbacks suppressed
Jan 31 14:01:40 kernel: net_ratelimit: 169601 callbacks suppressed
Jan 31 14:01:45 kernel: net_ratelimit: 171405 callbacks suppressed
Jan 31 14:01:50 kernel: net_ratelimit: 172444 callbacks suppressed
Jan 31 14:01:55 kernel: net_ratelimit: 182709 callbacks suppressed
*** Resume PC2 again
Jan 31 14:02:00 kernel: eth4 (Ext switch port: 3) (Logical Port: 11) Link UP 1000 mbps full duplex
*** PC1 has network connectivity again
Jan 31 14:02:00 kernel: net_ratelimit: 176069 callbacks suppressed
Jan 31 14:02:05 kernel: net_ratelimit: 88224 callbacks suppressed
Jan 31 14:02:23 kernel: wfd_unregisterdevice Successfully unregistered ifidx 1 wfd_idx 0
Jan 31 14:02:23 kernel: wfd_unregisterdevice Successfully unregistered ifidx 1 wfd_idx 1
Jan 31 14:02:26 kernel: wfd_registerdevice Successfully registered dev wds0.0.3 ifidx 1 wfd_idx 0
Jan 31 14:02:26 kernel: Register interface [wds0.0.3] MAC: b0:6e:bf:xx:xx:xx
Jan 31 14:02:27 kernel: wfd_registerdevice Successfully registered dev wds1.0.3 ifidx 1 wfd_idx 1
Jan 31 14:02:27 kernel: Register interface [wds1.0.3] MAC: b0:6e:bf:xx:xx:xx
Jan 31 14:02:30 roamast: eth5: add client [f0:03:8c:xx:xx:xx] to monitor list
*** Standby PC2 again
Jan 31 14:03:45 kernel: eth4 (Ext switch port: 3) (Logical Port: 11) Link DOWN.
*** PC1 loses network connection again. waiting
Jan 31 14:04:10 kernel: net_ratelimit: 126559 callbacks suppressed
Jan 31 14:04:15 kernel: net_ratelimit: 176078 callbacks suppressed
Jan 31 14:04:20 kernel: net_ratelimit: 177879 callbacks suppressed
Jan 31 14:04:25 kernel: net_ratelimit: 166737 callbacks suppressed
Jan 31 14:04:30 kernel: net_ratelimit: 179658 callbacks suppressed
Jan 31 14:04:35 kernel: net_ratelimit: 166097 callbacks suppressed
Jan 31 14:04:40 kernel: net_ratelimit: 165306 callbacks suppressed
Jan 31 14:04:45 kernel: net_ratelimit: 165562 callbacks suppressed
Jan 31 14:04:50 kernel: net_ratelimit: 167024 callbacks suppressed
Jan 31 14:04:55 kernel: net_ratelimit: 165358 callbacks suppressed
Jan 31 14:05:00 kernel: net_ratelimit: 176627 callbacks suppressed
Jan 31 14:05:05 kernel: net_ratelimit: 176284 callbacks suppressed
Jan 31 14:05:10 kernel: net_ratelimit: 164857 callbacks suppressed
Jan 31 14:05:15 kernel: net_ratelimit: 165861 callbacks suppressed
Jan 31 14:05:20 kernel: net_ratelimit: 165095 callbacks suppressed
Jan 31 14:05:25 kernel: net_ratelimit: 182026 callbacks suppressed
Jan 31 14:05:30 kernel: net_ratelimit: 174007 callbacks suppressed
Jan 31 14:05:35 kernel: net_ratelimit: 178554 callbacks suppressed
Jan 31 14:05:40 kernel: net_ratelimit: 166444 callbacks suppressed
Jan 31 14:05:45 kernel: net_ratelimit: 181106 callbacks suppressed
Jan 31 14:05:50 kernel: net_ratelimit: 170976 callbacks suppressed
Jan 31 14:05:55 kernel: net_ratelimit: 176695 callbacks suppressed
Jan 31 14:06:00 kernel: net_ratelimit: 166212 callbacks suppressed
Jan 31 14:06:00 roamast: eth5: add client [50:c7:bf:xx:xx:xx] to monitor list
Jan 31 14:06:05 kernel: net_ratelimit: 168028 callbacks suppressed
Jan 31 14:06:10 kernel: net_ratelimit: 176777 callbacks suppressed
Jan 31 14:06:15 kernel: net_ratelimit: 165737 callbacks suppressed
Jan 31 14:06:20 kernel: net_ratelimit: 172389 callbacks suppressed
Jan 31 14:06:25 kernel: net_ratelimit: 167916 callbacks suppressed
Jan 31 14:06:30 kernel: net_ratelimit: 164802 callbacks suppressed
Jan 31 14:06:35 kernel: net_ratelimit: 163943 callbacks suppressed
Jan 31 14:06:40 kernel: net_ratelimit: 176568 callbacks suppressed
Jan 31 14:06:45 kernel: net_ratelimit: 167171 callbacks suppressed
Jan 31 14:06:50 kernel: net_ratelimit: 170062 callbacks suppressed
Jan 31 14:06:55 kernel: net_ratelimit: 173788 callbacks suppressed
Jan 31 14:07:00 kernel: net_ratelimit: 170299 callbacks suppressed
Jan 31 14:07:05 kernel: net_ratelimit: 170278 callbacks suppressed
Jan 31 14:07:10 kernel: net_ratelimit: 182016 callbacks suppressed
Jan 31 14:07:15 kernel: net_ratelimit: 173293 callbacks suppressed
Jan 31 14:07:20 kernel: net_ratelimit: 174585 callbacks suppressed
Jan 31 14:07:25 kernel: net_ratelimit: 166974 callbacks suppressed
Jan 31 14:07:30 kernel: net_ratelimit: 165805 callbacks suppressed
Jan 31 14:07:36 kernel: net_ratelimit: 166469 callbacks suppressed
Jan 31 14:07:41 kernel: net_ratelimit: 165415 callbacks suppressed
Jan 31 14:07:46 kernel: net_ratelimit: 167554 callbacks suppressed
Jan 31 14:07:51 kernel: net_ratelimit: 166086 callbacks suppressed
Jan 31 14:07:56 kernel: net_ratelimit: 167929 callbacks suppressed
Jan 31 14:08:01 kernel: net_ratelimit: 166486 callbacks suppressed
Jan 31 14:08:06 kernel: net_ratelimit: 159741 callbacks suppressed
Jan 31 14:08:11 kernel: net_ratelimit: 167145 callbacks suppressed
Jan 31 14:08:16 kernel: net_ratelimit: 184738 callbacks suppressed
Jan 31 14:08:21 kernel: net_ratelimit: 174394 callbacks suppressed
Jan 31 14:08:26 kernel: net_ratelimit: 165252 callbacks suppressed
Jan 31 14:08:31 kernel: net_ratelimit: 170875 callbacks suppressed
Jan 31 14:08:36 kernel: net_ratelimit: 179260 callbacks suppressed
Jan 31 14:08:41 kernel: net_ratelimit: 178970 callbacks suppressed
Jan 31 14:08:56 kernel: wfd_unregisterdevice Successfully unregistered ifidx 1 wfd_idx 0
Jan 31 14:08:56 kernel: wfd_unregisterdevice Successfully unregistered ifidx 1 wfd_idx 1
Jan 31 14:09:00 kernel: wfd_registerdevice Successfully registered dev wds0.0.3 ifidx 1 wfd_idx 0
Jan 31 14:09:00 kernel: Register interface [wds0.0.3] MAC: b0:6e:bf:xx:xx:xx
Jan 31 14:09:00 kernel: wfd_registerdevice Successfully registered dev wds1.0.3 ifidx 1 wfd_idx 1
Jan 31 14:09:00 kernel: Register interface [wds1.0.3] MAC: b0:6e:bf:xx:xx:xx
Jan 31 14:09:06 kernel: net_ratelimit: 103843 callbacks suppressed
*** PC1 regains network connection

 

[scope666]

Do you have a large number of clients? Maybe the list gets full (size wise) immediately. I had about 22 clients and have some of the names kind-of long. I just went in to shorten the more verbose ones.
If you use the nvram commands I show above, the list will get rebuilt immediately after the reboot. The default name is sent from the client. If there are lots of client devices, maybe temporarily reduce the number until you can shorten the longer strings.

I don't believe it's corruption. I have done a nvram get custom_clientlist and manually inspected the entries, everything is correct. There is a limit of about 1000 characters in that field, so I tend to only add the smaller embedded devices that don't report their hostnames well.

I never ran into this limit on Merlin, I'm wondering if he allowed more than a 1000 characters for custom_clientlist.

Also there seems to be a separate issue of the service that shows the clients getting hung up, and showing old data. (you can tell because normally your total client number will change every few seconds. When it gets stuck this number won't change again)

This is why I was asking if anyone knew the culprit service. It would be great to be able to restart just that service without a full reboot.

 

[RMerlin]

I never ran into this limit on Merlin, I'm wondering if he allowed more than a 1000 characters for custom_clientlist.

The 1000 chars limit on the RT-AC86U is enforced at the SDK level by Broadcom.

I know Asus recently went around that limitation by moving some settings to JFFS (like dhcp_staticlist). I don't remember if custom_clientlist was one of them.

 

[Ronald Schwerer]

The 1000 chars limit on the RT-AC86U is enforced at the SDK level by Broadcom.

I know Asus recently went around that limitation by moving some settings to JFFS (like dhcp_staticlist). I don't remember if custom_clientlist was one of them.

Checking my /jffs, I see only nmp_client_list. While this looks similar to the custom_clientlist, it has the original client names. On mine, it is a little over 1K and doesn't even contain the MACs.
However, custom_clientlist is stored in /dev/mtd1 along with nvram.

 

[scope666]

Checking my /jffs, I see only nmp_client_list. While this looks similar to the custom_clientlist, it has the original client names. On mine, it is a little over 1K and doesn't even contain the MACs.
However, custom_clientlist is stored in /dev/mtd1 along with nvram.

Nice find. I see similar data in nmp_cl_json.js I'm guessing we probably can't edit these to get around the limitations of custom_clientlist


"E0:4F:03:4A:8C:0B":{
"type":5,
"mac":"E0:4F:03:4A:8C:0B",
"name":"ANTSCAM-0000-TNPUSAC-397610-GLX",
"vendor":"Hewlett-Packard"


Edit: nmp_client_list does contain the MACs, but without the colons.

 

[GrinningShark]

Do you have a large number of clients? Maybe the list gets full (size wise) immediately. I had about 22 clients and have some of the names kind-of long. I just went in to shorten the more verbose ones.
If you use the nvram commands I show above, the list will get rebuilt immediately after the reboot. The default name is sent from the client. If there are lots of client devices, maybe temporarily reduce the number until you can shorten the longer strings.

Funny part is that 22 clients is not that much for a router like this. I realize that space may be finite, but the client list should not just stop working even when I do not customize anything. A reboot fixes it, but then it just dies again. It should be able to easily handle this. Oh well, I will mess around with some of the suggestions.

 

[scope666]

Funny part is that 22 clients is not that much for a router like this. I realize that space may be finite, but the client list should not just stop working even when I do not customize anything. A reboot fixes it, but then it just dies again. It should be able to easily handle this. Oh well, I will mess around with some of the suggestions.

I agree 100% ... based on Merlin's comment I don't see why they can't move it to /jffs. As it stands now there's not enough space to hold all of the devices on my network, so I can only use custom names / icons for a smaller percentage of the network.

 

[Stephen Harrington]

I agree 100% ... based on Merlin's comment I don't see why they can't move it to /jffs. As it stands now there's not enough space to hold all of the devices on my network, so I can only use custom names / icons for a smaller percentage of the network.

+1 to the above. It's ridiculous that there is this limitation in what is supposed to be one of their flagship models. Its a definite step backwards.
How can we apply some serious pressure to Asus to get it fixed?

StephenH

 

[Sendil K Krishna]

Option 1: disabled roaming assistant
Option 2: adjust threshold from -50 to others.

我從使用 Tapatalk 的 ASUS_Z012DA 發送

I understand that increasing the threshold (option 2) might prevent the client from getting kicked off prematurely. my question is, is setting the threshold t0 -90 dBm (typical noise floor) achieve the same as option 1 or does option 1 have other implications?

 

[RandomName23]

+1 to the above. It's ridiculous that there is this limitation in what is supposed to be one of their flagship models. Its a definite step backwards.
How can we apply some serious pressure to Asus to get it fixed?

StephenH

Are there any models that don’t have this limitation?

 

[y2jdmbfan]

Any idea why the 86U is defaulting to channels 10 and 151? It's totally overlapping other networks. Looks like I will have to manually set them.

 

[RMerlin]

How can we apply some serious pressure to Asus to get it fixed?

Apply pressure to Broadcom instead. They're the ones that came up with this ridiculous limit, and Asus is forced to work around it whenever they can.

Sometimes, I wonder if Broadcom didn't deliberately sabotage their own HND platform, considering the number of stupid design decisions on their part: going for an LTS kernel without keeping it up-to-date, imposing 1000 chars limitations to nvram values and enforcing it within closed source components tied to wireless (!), implementing a kernel module that reserves 13% of the total available RAM for its own use, and the list goes on...

 

[scope666]

Apply pressure to Broadcom instead. They're the ones that came up with this ridiculous limit, and Asus is forced to work around it whenever they can.

Sometimes, I wonder if Broadcom didn't deliberately sabotage their own HND platform, considering the number of stupid design decisions on their part: going for an LTS kernel without keeping it up-to-date, imposing 1000 chars limitations to nvram values and enforcing it within closed source components tied to wireless (!), implementing a kernel module that reserves 13% of the total available RAM for its own use, and the list goes on...

I wonder if Asus could employ a work around like you mentioned they did with dhcp_staticlist.

custom_clientlist is an important one considering how many devices consume an IP address these days. I'm at around 30 clients and 1000 characters is nowhere near large enough to accommodate them all, even with incredibly short names.

 

[RMerlin]

I wonder if Asus could employ a work around like you mentioned they did with dhcp_staticlist.

Technically I don't see why not. Try sending them a request for it through Router Feedback.

 

[Sendil K Krishna]

use command line, set log level to 7, original is 6, and reboot the node/router using reboot:

nvram set log_level=7
nvram commit
reboot

Thanks!

I have a fairly smallish home but I was trying to use aimesh to improve backyard and frontyard coverage for security cameras. Apparently, a side-effect of having closely spaced nodes seems to be causing stationary devices (TV, chromecast etc) drop wifi sporadically.

I have made the following additional changes now -- Airtime Fairness & Universal beam-forming - Disabled and Roaming assistant threshold at -65/-70 dBm.

hopefully, I should be able to monitor if they are getting kicked off by "roam assist" feature unnecessarily.

 

[Stephen Harrington]

Apply pressure to Broadcom instead. They're the ones that came up with this ridiculous limit, and Asus is forced to work around it whenever they can.

Sometimes, I wonder if Broadcom didn't deliberately sabotage their own HND platform, considering the number of stupid design decisions on their part: going for an LTS kernel without keeping it up-to-date, imposing 1000 chars limitations to nvram values and enforcing it within closed source components tied to wireless (!), implementing a kernel module that reserves 13% of the total available RAM for its own use, and the list goes on...

Nice thought but I'm not Broadcom' s customer, I'm Asus's , and besides, where would I even start as an individual?

Maybe @arthurlien will read this and take up the cause?
I will submit a request through Router Feedback though about making custom_clientlist larger and/or putting it into NVRAM.
May I ask all affected to do the same? Maybe if enough of us do it we may see a result eventually?
Seems such a stupid limitation, my AC68U does it better and its design is getting pretty long in the tooth ...

StephenH

 

[RMerlin]

Nice thought but I'm not Broadcom' s customer, I'm Asus's , and besides, where would I even start as an individual?

My point was this limitation was put in place by Broadcom, not by Asus.

 

[Stephen Harrington]

My point was this limitation was put in place by Broadcom, not by Asus.

Yes @RMerlin I get it!

I've read your explanations about the platform (and as always, appreciate your incredible work both on your firmware and also generally here in the SNB Forums).
Want to run your firmware of course for selective routing, and then run into the silliness of custom_clientlist being limited to 1000 chars.
Particularly annoying glitch in my plan to upgrade from an AC68U to a AC86U - to get some more VPN horsepower and future-proof myself a bit.
I have around 5o+ devices, which I think is not atypical nowadays, and custom names makes it all a LOT easier - hard to go back when you've had it!

Thanks,

StephenH