AiMesh randomly disconnecting - "MISC: wifi upstream is connected, and disconnected from CAP."

[bibikalka]

I am running a 2 router AiMesh, AC86U and AC68U, both on Merlin 386.14_2. (added later) The backbone is the 5Ghz band, and there is no wired connection between the devices.

Pretty randomly AC68U node disconnects, and then reconnects. I reboot AC68U daily, so it's fresh. I tried another AC68U device, same issue.

The key telltale sign is this message on the AC68U node (I have removed the same message during the node reboot):

admin@RT-AC68U-FF68:/tmp# grep MISC syslog.log |grep -v " 04:07"
May  2 08:19:13 MISC: wifi upstream is connected, and disconnected from CAP.
May  2 10:51:52 MISC: wifi upstream is connected, and disconnected from CAP.
May  2 11:36:10 MISC: wifi upstream is connected, and disconnected from CAP.
May  3 19:40:11 MISC: wifi upstream is connected, and disconnected from CAP.
May  3 21:58:25 MISC: wifi upstream is connected, and disconnected from CAP.
May  3 22:04:14 MISC: wifi upstream is connected, and disconnected from CAP.
May  4 12:08:16 MISC: wifi upstream is connected, and disconnected from CAP.
May  4 12:12:32 MISC: wifi upstream is connected, and disconnected from CAP.
May  4 12:43:30 MISC: wifi upstream is connected, and disconnected from CAP.
May  5 07:08:17 MISC: wifi upstream is connected, and disconnected from CAP.
May  5 07:14:16 MISC: wifi upstream is connected, and disconnected from CAP.
May  5 08:38:04 MISC: wifi upstream is connected, and disconnected from CAP.
May  5 10:23:39 MISC: wifi upstream is connected, and disconnected from CAP.

I vaguely recall this had been more stable in the past. But I am not sure - I did not track it much back then.

I guess AiMesh is proprietary - so is there some easy tracker which Merlin versions had which AiMesh versions?

 

[ramthrou]

Merlin 386.13 was the latest I was able to run reliably with AC86U as the main router in an Aimesh setting, and from my understanding I'm not alone. So stick to 386.13 if you need Merlin.

It's getting outdated tho, so your long term plan should be to upgrade the main node to an AX unit that can run 388.9, and then religate the wifi5 units to Aimesh-node status. When you do so, 388.7 is the latest I was able to run reliably there, 388.8 gave me Aimesh stability issues, and 388.9 I haven't tried yet (waiting for somebody to give me some reports here.)

Use Asuswrt (official fw, as opposed to Merlin) on the nodes.

 

[bibikalka]

Merlin 386.13 was the latest I was able to run reliably with AC86U as the main router in an Aimesh setting, and from my understanding I'm not alone. So stick to 386.13 if you need Merlin.
...

I mean, this issue was mentioned in the past, as in here - Link

How confident are you that there was no such issue on 386.13 ?

 

[bbunge]

With these old EOL routers it may be best to switch to Asus firmware.

 

[ramthrou]

I mean, this issue was mentioned in the past, as in here - Link

How confident are you that there was no such issue on 386.13 ?

I didn't say that .14 was the first firmware with disconnection issues, FAR from it (I've travelled to hell and back with AC86U instability, the firmware stability record is atrocious, so this is some seriously hard-earned knowledge I'm dropping here, I've seen things you would't believe..).

What I said was that .13 was the last I was able to get to work reliably. In fact, I'm running it as we speak as a temporary stand-in for my AX router because of reasons. If .13 also fails on long-term stability, I'd try .11.

With these old EOL routers it may be best to switch to Asus firmware.

For all nodes, new or old, official firmware is recommended, as it has a better track record of Aimesh stability. There is nothing to gain from running Merlin on the nodes, as the extra functionality runs on the main-router.

 

[bibikalka]

...
What I said was that .13 was the last I was able to get to work reliably. In fact, I'm running it as we speak as a temporary stand-in for my AX router because of reasons. If .13 also fails on long-term stability, I'd try .11.
...
For all nodes, new or old, official firmware is recommended, as it has a better track record of Aimesh stability. There is nothing to gain from running Merlin on the nodes, as the extra functionality runs on the main-router.

Not sure which issues you were having - I have .14_2 running nicely on my AC86u. It's auto-rebooted nightly - so there is that. Once a month I get stuck "nvram get" command, so I notice flakiness and may have to force reboot.

AiMesh is mostly OK, occasionally it drops for a few moments, but then clients grab the more remote main router. The bigger deal are the Asus features like parental control (AiProtection) which I need to run and which mini-crash the main router from time to time.

I guess I'd like reports of someone with a functioning AiMesh who is not having these ""MISC: wifi upstream is ..." problems at all.

 

[ramthrou]

Not sure which issues you were having - I have .14_2 running nicely on my AC86u. It's auto-rebooted nightly - so there is that. Once a month I get stuck "nvram get" command, so I notice flakiness and may have to force reboot.

AiMesh is mostly OK, occasionally it drops for a few moments, but then clients grab the more remote main router. The bigger deal are the Asus features like parental control (AiProtection) which I need to run and which mini-crash the main router from time to time.

I guess I'd like reports of someone with a functioning AiMesh who is not having these ""MISC: wifi upstream is ..." problems at all.

You're contradicting yourself quite a bit: first you start a post because of disconnection issues, now you say it's "running nicely" before you outline why its not running nicely again. Then you challenge my assertion that .13 is running better for me for no reason instead of trying it out yourself. What do you want, somebody to tell you that your .14 is running nicely, or for it to actually start running nicely out of the blue so that you can say it ran nicely all along, even if you started a thread because it was not running nicely...

The issues I was having before was the same as you report: disconnection issues. Plus I had issues with a crashing web-ui which required manual restart or via the command line through ssh.

This is a while ago, so I from what I recall it was .13 that ran better. .14 was a disaster. But in case I forget something, I included a tip to try .11 in case that also fails, because I experienced quite a lot of issues around that time which also might have had to do with the node firmware versions. I should clarify that the .13 I'm running now temporarily while my AX router is offline is as a main router with AP's, not Aimesh, so it may be that it was .11 I ran about a year ago when my AC86 was in daily operation as the main node in my Aimesh network (this was another location.

So in short, try .13, and if that doesn't resolve the issues with disconnections, try .11.
If that doesn't work either, try the latest official FW.

 

[bibikalka]

You're contradicting yourself quite a bit: first you start a post because of disconnection issues, now you say it's "running nicely" before you outline why its not running nicely again. Then you challenge my assertion that .13 is running better for me for no reason instead of trying it out yourself. What do you want, somebody to tell you that your .14 is running nicely, or for it to actually start running nicely out of the blue so that you can say it ran nicely all along, even if you started a thread because it was not running nicely...
...
So in short, try .13, and if that doesn't resolve the issues with disconnections, try .11.
If that doesn't work either, try the latest official FW.

Well, I guess it's a matter of degree? It occasionally disconnects, seems like 3-4 times a day for a couple of minutes, and requires no action from me to get it back working But otherwise the system works reasonably well. I just wanted to perhaps figure out if somebody has such a setup rock solid.

Chasing version downgrades is always a possibility, but I'd like more evidence that somebody has it on a lower version, and it has no such issues. At least, if you produced some logs to indicate stability it'd be more pervasive.

P.S. I forgot to mention that I am using 5Ghz band as the backbone between the routers, and there is no wired connection.

 

[ramthrou]

It occasionally disconnects, seems like 3-4 times a day for a couple of minutes, and requires no action from me to get it back working But otherwise the system works reasonably well. I just wanted to perhaps figure out if somebody has such a setup rock solid.

And your answer is yes, somebody does.

Chasing version downgrades is always a possibility, but I'd like more evidence that somebody has it on a lower version, and it has no such issues. At least, if you produced some logs to indicate stability it'd be more pervasive.

That's unlikely to happen for a few reasons, one of which Merlin has abandoned the 386 firmware branch and people have started upgrading their main nodes to AX routers capable of running the 388 branch or firmware (which has its headaches too, read my comment history).

How about this suggestion: you implement the advice i gave you by rolling back to .13 for one day and see if there are any improvements, and then if not, roll back to .11 and see then.

Alternatively, if you want to cut right to the chase and don't mind loosing whatever Merlin features you use; install the official FW for one day and see if that confirms that your issue is firmware related, in which case you can start rolling back the Merlin versions until the same stability is achieved there.

And don't forget to report back your results in case it gets resolved.

Sounds like a plan?

 

[bibikalka]

...

Sounds like a plan?

It's a plan, but a poor one

Basically, you offered nothing but very generic trouble shooting. I can do that if I need to myself. You provide no specifics that there is any reasonable expectation that a different version might change something. So the (expected success)/(time spent) ratio is very poor. I appreciate your effort to help - but it's just too generic.

I can live with the issue for now, but if somebody else stumbles on this, and gets some more specific ideas and observations given that their setup is either doing or not doing the same behavior - that'd be something.

It just may be that WiFi AiMesh backbones are always doing that, and most people are running wires ones so they don't see this behavior.

 

[ramthrou]

It's a plan, but a poor one

Basically, you offered nothing but very generic trouble shooting. I can do that if I need to myself.

In that case, you should have done so before starting this thread, or at least specify that your primary goal is not to solve the problem, but stay on the newest firmware.

You provide no specifics that there is any reasonable expectation that a different version might change something.

With respect, if you had some experience with Asus routers and the lack of QC on their firmwares, you'd know that trying to resolve an issue before you have checked whether it was introduced with a firmware upgrade, is a fool errand, and a very good recipe for wasting precious life-time.

It just may be that WiFi AiMesh backbones are always doing that, and most people are running wires ones so they don't see this behavior.

No, that's not the case. It is not normal. Should go without saying, and now I'm starting to suspect that this is purely an ego thing with you, not wanting to take advice because of embarrassment because you think you look bad for not figuring this out yourself. Not that I think you're right about that.

 

[bibikalka]

Did you guys ever get this figured out? Not sure why this is all of the sudden happening to my ecosystem.

There is more of this stuff happening. Are you using 5Ghz backbone?

 

[bibikalka]

@ramthrou
As mostly expected, the downgrade did nothing. Disconnects continue. Vague advice rarely (if ever!) works ...

I downgraded the mesh node to 386.13_0, and the main AC86u node to 386.13_2. Version 386.14_2 on the headnode would not play properly with the mesh node at .13, so I head to downgrade both. No improvement.

admin@RT-AC68U-FF68:/jffs# grep MISC syslog.log-1 |grep -v " 04:07"
May 10 14:47:04 MISC: wifi upstream is connected, and disconnected from CAP.
May 10 14:52:51 MISC: wifi upstream is connected, and disconnected from CAP.
May 10 18:22:37 MISC: wifi upstream is connected, and disconnected from CAP.
May 10 18:42:22 MISC: wifi upstream is connected, and disconnected from CAP.
May 11 08:31:55 MISC: wifi upstream is connected, and disconnected from CAP.
May 11 08:37:43 MISC: wifi upstream is connected, and disconnected from CAP.
May 11 11:22:07 MISC: wifi upstream is connected, and disconnected from CAP.
May 11 11:27:53 MISC: wifi upstream is connected, and disconnected from CAP.
May 12 13:17:26 MISC: wifi upstream is connected, and disconnected from CAP.
May 12 13:23:13 MISC: wifi upstream is connected, and disconnected from CAP.
May 12 20:52:06 MISC: wifi upstream is connected, and disconnected from CAP.
May 13 07:22:30 MISC: wifi upstream is connected, and disconnected from CAP.
May 15 00:27:21 MISC: wifi upstream is connected, and disconnected from CAP.
May 15 00:33:09 MISC: wifi upstream is connected, and disconnected from CAP.
May 16 04:05:46 MISC: wifi upstream is connected, and disconnected from CAP.
May 16 11:10:25 MISC: wifi upstream is connected, and disconnected from CAP.
May 16 14:28:21 MISC: wifi upstream is connected, and disconnected from CAP.
May 16 14:34:05 MISC: wifi upstream is connected, and disconnected from CAP.
May 16 19:56:29 MISC: wifi upstream is connected, and disconnected from CAP.
May 17 11:24:50 MISC: wifi upstream is connected, and disconnected from CAP.
May 17 20:07:56 MISC: wifi upstream is connected, and disconnected from CAP.
May 17 20:13:43 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 01:22:32 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 01:28:18 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 13:47:32 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 13:53:19 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 17:57:46 MISC: wifi upstream is connected, and disconnected from CAP.
May 19 18:03:33 MISC: wifi upstream is connected, and disconnected from CAP.

 

[ramthrou]

@ramthrou
As mostly expected, the downgrade did nothing. Disconnects continue. Vague advice rarely (if ever!) works ...

I downgraded the mesh node to 386.13_0, and the main AC86u node to 386.13_2. Version 386.14_2 on the headnode would not play properly with the mesh node at .13, so I head to downgrade both. No improvement.

That is the expected result, because you didn’t follow my instructions carefully. My advice was very specific:

1. Firmware versions:
   • ONLY the main node should be running Merlin firmware. All other AiMesh nodes should be on the latest official AsusWRT firmware.
   • Try flashing the AC86U’s Merlin firmware back to version .13. If that doesn’t work, try .11.
2. If that doesn’t work (or you might just jump to this step right away and possibly save time):
   • Flash the main node to the latest official firmware (or the last official firmware you know was stable for your setup, we're doing a process of elimination at this point) just like you did for all the nodes. Roll back if necessary.
   • If this doesn't solve your stability issues, the problem is either hardware failure (which is very rare, despite what some forum members might claim). More often, it’s due to a problematic setting in your environment—setting the 2.4 GHz bandwidth to 20 MHz is a common fix.
   • If official firmware is stable, you can go back to trying try different Merlin versions. Note that the AC86U is known for being especially picky about firmware versions, so some experimentation may be needed.

 

[bibikalka]

That is the expected result, because you didn’t follow my instructions carefully.

Are you actually for real???

For the sake of everyone's time, please stop giving any and all advice from now on.

 

[ramthrou]

Are you actually for real???

For the sake of everyone's time, please stop giving any and all advice from now on.

Do you have a brain issue/processing impediment? You nonchalantly complain about my advice not working, and then you proceed to list what you did, which turns out to have nothing to do with what I recommended. I point out that no fix is the expected result.

Wtf is even your argument, and why are you still bitching like you have some sort of factual highground?

My new advice: throw your router in the garbage and ask your dad to order a router and mesh point from your ISP that is remotely serviced by them. You are clearly not smart/mature enough to handle this stuff.