RT-AC88U Drops/Flakey 5Ghz when set as AiMesh Node

[JDB]

Hi All,

So I've been running AiMesh with a pair of AC5300 (one as AP/AiMesh Router and the other as a node) for about 3 months now, and it's great.
I recently added an AC88U in as a second node, see this thread for a bit more background on why/my architecture (also see my signature for my setup info) - https://www.snbforums.com/threads/my-new-rt-ac86u-and-some-qs-issues.50058/

It's been running OK for a couple of days, but then yesterday I noticed my iPhone WiFi was not full strength despite having been sat 5 ft from the 88U for 30 mins, so I looked into what was going on.
It was no long broadcasting 5Ghz at all, despite having the LED on, now I had just mounted it on the wall properly (was wedged on top of the TV for a couple of days before I committed to it's location) and when doing so I had knocked the WiFi On/Off button, so I gave it the benefit of the doubt that maybe something had gone fooey because of that. Cold booted it, 5Ghz came back for a few mins, then dropped again.

It's worth noting that prior to being a Node, this was my main router (running Merlin) giving me rock solid 5Ghz only a few days ago. Since then was to flashed to the latest stock f/w, reset to factory and setup as a node.

So at first I was thinking it was overheating, but given its only 18C in the room it's in and mounted to a cold brick wall I discounted that pretty quickly.

Decided to factory reset again, re-add as a node, and hey - it's working again! Still I'm hoping it just wasn't happy with the fact I'd turned it's WiFi off and on and maybe something in the NVRAM was set wrong as a result and persisting over a reboot.

So next day (today), this time my phone is dropping on and off the WiFi (sat in the same spot). Run some continuous WiFi scans and can see the 5Ghz is again playing up. It's going up and down, hopping channels and all sorts. So I reboot it, comes back doing the same thing. It was also clear the AiMesh was trying to roam the phone to the 88U and sometimes the BSSID was not there (or maybe not on the channel it was supposed to be), hence the phone would drop WiFi.
So now I decided to reboot the AP/Router and the other Node, so all AiMesh devices have been cycled, and hey, after that everything is good again!

So I'm thinking the problem is maybe the master AP/Router is telling the node to do these 5Ghz gymnastics.

In the couple of days good running I had, I did a bunch of tests walking around monitoring signal and checking roaming handover speeds etc. I noted that one time I did it, the 88U was using the 5Ghz upper channel and the next day it had switched to the lower channel (in both cases was correctly sharing the channel of the AP/Router, obviously the 5300 has 2x 5Ghz radios). I didn't think much of it, just figured it would use which ever it saw was least congested and so occasionally would swap between upper and lower. Maybe it's this choice of 2x 5Ghz channels that is getting it all upset sometimes?

Per signature, I have not updated the fw to the vary lasted versions just released due to known issues with 5300 and the 2nd 5Ghz band being disabled.


I know there are various threads about AiMesh and it breaking with 3+ nodes and also about the 88U dropping 5Ghz when in normal router mode (thought I doubt the latter one is relevant to my issue), so hoping maybe someone somewhere has experienced similar and got an idea of what's happening. All the threads I've read so far haven't drawn any solid conclusions yet.


If you've got here, thanks for reading, I know it's long, but wanted to share all the facts!

Cheers!

 

[JDB]

@Krusty tagging you as although your issues aren't exactly the same, they are as close as I can find!
@HuskyHerder tagging you as you also have a large AiMesh so might have insights!

Worth noting, all 3 of us have HomeKit and a significant number of smart devices by the sounds of it.

It's not mentioned in my signature, but I run an Apple Time Capsule 3TB as a 2.4Ghz only AP as a separate SSID to the AiMesh, it has attached to it the various older and wireless b/g only devices, so my AiMesh 2.4Ghz has primarily newer wireless N compatible devices only.
Reasons are I know mixing old with new doesn't do the router/network any favours with efficiency, hence the existence of options like Airtime Fairness (which many people have to disable inc myself for some devices to work).

 

[Krusty]

@Krusty tagging you as although your issues aren't exactly the same, they are as close as I can find!
@HuskyHerder tagging you as you also have a large AiMesh so might have insights!

Worth noting, all 3 of us have HomeKit and a significant number of smart devices by the sounds of it.

It's not mentioned in my signature, but I run an Apple Time Capsule 3TB as a 2.4Ghz only AP as a separate SSID to the AiMesh, it has attached to it the various older and wireless b/g only devices, so my AiMesh 2.4Ghz has primarily newer wireless N compatible devices only.
Reasons are I know mixing old with new doesn't do the router/network any favours with efficiency, hence the existence of options like Airtime Fairness (which many people have to disable inc myself for some devices to work).

Hey JDB... I should probably have mentioned that I am running with 3 different SSIDs for 5/5/2.4 and no smart connect... sounds like that is another variable you may have.

Note also that it is a router + 3 nodes that breaks for me, not router + 2 nodes... though in all fairness as per other thread I do have occasional problems that can be solved by taking 2.4Ghz interface down on the main router even with two - my suspicion is that the whole issue may be dependent on the relative placement of the routers for Wifi between each other (even though I'm using wired backhaul).

I'm gonna try to re-add the 3rd node today

 

[HuskyHerder]

As for me, and my setup, see signature. Thanks for the assistance @JDB

2.4G channel Devices:

1, ecobee
2, Apple Watches, they never connect on 5G even with the ways on the net to force them. Thats not an issue so its on the back burner.
1, Phillips Hue Hub with 37 bulbs
5, iDevices smart switches * wall wart kind
1, iDevice Wall switch.

HomeKit setup:
4 Apple TVs as hubs, 1 ATV4K as primary and 3 ATV 4 (non k) in standby mode.

Phillips Hue:
The single most relevant thing for me as to keeping my hue bulbs working was to change the control channel between the Hue Hub and the Bulbs. It was set at 20 and I force changed it to 25. I did some hunting at the Phillips forums and this topic came up after finding it. It has helped immensely. https://www.metageek.com/training/resources/zigbee-wifi-coexistence.html



Routers changes:

No Smart connect with the 2.4 channels and the 5G channels. Smart connect was only used in the older 3 channel version for the 5300 just between the 2, 5G channels. Now it's not used at all. Thats another issue.

* Everything is default unless mentioned below.

Using WPA2 with very strong encryption. *Including a few special characters which both Apple and ASUS hate at times.
2.4G RSSI: -48 * My specific setting not for everyone.
Airtime Fairness: - disable
Explicit Beam Forming: - disable
Universal Beam Forming: - disable
Power level: Balanced

Notes:

Also make sure "Enable IGMP Snooping" is disabled. I experienced huge issues if it's enabled, flaky response, no response, severe lag etc.
I also saw a post to change the Modulation Scheme from 11 to 9 if the Router supports it. I also experienced very poor performance with this. It took all of 10 minutes and I switched it back to 11.

Special Environmental considerations.

1950's home with plaster wall and wire mesh backing.
This means the signal has to go through plaster 1/2" - 3/4" thick - Wire mesh, plaster that bleed through the back of the mesh, 3" - 4" dead space, plaster that bleed through the back of the mesh, Wire mesh, plaster 1/2" - 3/4" thick before reaching the next room.
The home has foil insulation backing around the entire home and 1 interior wall. * That interior wall use to be an exterior wall before an addition was added.



What does this mean. In room (a), the signal is a 5 /5 the signal in the next room (b) from the same AP is a 1/5 or less. Usually totally unusable. Occasionally it will be but is so degraded it may as well be non existent.

Solution for me: 1 Node in the outside corners of the home with antennas pointed to the interor, and 1 near the direct center of the home. Makes life a lot easier for me. This diagram is a rough approximation of my home, and the Router & Node setup. It's just a hastily constructed representation.
There are some closets and 2 hallways, not show which increase the amount of penetration required. Instead of passing through one plaster cross section it requires 2. Because of the opposite wall as in the case of the closet. This further constrains the WiFi signals. The Brown box is a old brick chimney.



Other considerations:

My OCD compels me, to use static IPs on what I can. Basically everything in my home has a static IP.
I have used everything in AiProtection, enabled with no noticeable affect. Neither good nor bad.

I also removed an Arlo wireless camera system, which insisted on setting up its own WiFi network along side my own. However this still had no where near the impact to my system usage as when I changed the Hue control channel as indicated above. That Hue control channel change was the single greatest improvement to my system hands down. *I also think using mesh had some benefit as it uses the same channel and now I don't have competing channels all over the house as when I was strictly using AP's.

I think I covered my situation, if there is anything you think I missed give me a shout.

 

[JDB]

Wow, thanks!

I had the issue again today, interestingly after my power tripped as a result of my combi oven blowing its element!
Again a reboot of all AP/Nodes fixed it, but it’s very annoying that it can happen with a power restoration!!

I have set a scheduled reboot of all AP/Nodes for 4am so will see if it helps or makes it worse each morning when I get up the next few days.


Sent from my iPhone using Tapatalk

 

[HuskyHerder]

Oh I have not noticed any issues with the 5G I’ll keep an eye on mine, with the reports above and elsewhere here in the forums.

Edit:

1, other thing I did was relocate my Hue hub from within with my entertainment center to a location 15' from any Node, and really high up, next to the ceiling.
Previously I had used it on a shelf approximately 3' from the top of the wall, and had less than desireable results. Moving it higher also helped out.

 

[HuskyHerder]

Here are some ideas of why I need more than one or two nodes .

With my Full AiMesh Setup on 5G
AiMesh system
Several Nodes around the home
A
B
C
D
Node E is a detached Garage, about 50-75 from the home. During the Sumer months were outside constantly and that node allows easy migration from the home. It's also keeping several remote IoT devices connected.

Node C is my very best placement option, and it looks like I should be able to really cut down on my nodes. In reality though, I found the kids iOS devices were always dropping off WiFi and reverting to LTE. Causing data overages. I was always constantly going into the kids rooms and checking the WiFi which looked good, but the phones would drop off especially during the evening hours.
There were no time limits on there devices at that time, so there could only be environmental concerns. It was easily reproducible even with my iPhone. I would notice my phone dropped the WiFi after spending just a few minutes in their rooms.

Between the 1st and the 2nd the coverage looks almost identical. But again coverage would fail. Though between these 2 specific nodes it was for the most part stable.

*This coverage failure was not just ASUS products, it was documented by me, when I still used Apple AirPort Extreme / Airport Express units. I ended up having devices in the exact same locations as you see in my maps. Further more it was also the same way back when I still used Linksys products.

I have contemplated placing 2 nodes in the attic and seeing how things go. But here in the southern US the summer time weather is extremely hot, and the attics are beyond belief.
I know from experience it's not best to put them in the attic, and back when I installed home electronics, I absolutely refused to install things in the attic, due to device failures.

Additionally : The side of the home with 1 node in it. (lower right corner) Is the most recent addition to the home and it has normal sheetrock walls vs the Plaster on the original side of the home. This would explain the easier penetration of the WiFi signal over a larger are and needing only 1 node here.

 

[HuskyHerder]

The forum will not let me add more images to the above so this is a continuation.

2G coverage:

Checking the maps I made for the 2G network, they show ample coverage all around for the 2G band, off a single Node or Router, placed centrally in the home. The issue in the home is clearly the 5G and signal



BTW the circles on the Maps are the placements of the routers/nodes with the 5300 AiMesh Router being the large circle in the center and the 68U nodes the smaller circles.

 

[JDB]

Ok little update...

So the Powerline adaptor that the 88U was connected to completely lost the plot the other day. So disconnected it and went to wireless backhaul.
This resulted in 2 things;
1. Limited to 30-40Mbps (don’t really understand why as 5Ghz clients in the room get 300+ connected to the AiMesh router, even if it was using 2.4Ghz though I’d expect better!)
2. 88U no longer goes crazy!

So after a couple of days of slow but consistent usage my replacement powerline plug arrived.
Plugged it in fully expecting craziness to return within a few hours, but so far 2 days later we’re good.

So this leads me to think the powerline that failed was never right and was giving intermittent connection. The AiMesh network didn’t cope with this very well (really it should have a mechanism to stop using wired if it finds it inconsistent) and caused the odd behaviour.

So now I’m pondering if/how I can logistically get a cable from the conservatory up to the lounge on the next floor where the AiMesh router is! I think actually it’ll be quite easy to do an external run and in the same hole the satellite cable goes in anyway (especially as I’m technically already “outside” in the conservatory).


So if anyone is having odd stuff and are attempting wired backhaul, triple check your cabling is 100% as I bet a dodgy Cat5 would produce the same results.


Sent from my iPhone using Tapatalk

 

[HuskyHerder]

Glad to hear things are improving. Mind if I ask about the scheduled reboots? Are they a synced thing or do you manually set things up on the node via ssh ?

If I were closer to ya, I would offer to assist in running the cable. I did that almost 20 years ago now, for a living. I still dabble in it, for my own use and or my friends, when they need assistance.

I personally would invest in some exterior grade Cat cable. I am hesitant to say you could get away with normal cable, while it may be possible, I think you would prefer the results of exterior grade better and make it future proof. All the while making sure you didn't have to re do the work in as little as a few months to a year. People I worked for (customers) use to balk at the price difference for exterior grade. Usually it only amounted to $20 or so. * shakes head

If we are being totally honest here, I would often just eat that price, off my own commission. That was so, I didn't have to go back in 6 months to a year, wasting my time with a service call, fight with them over the cost of that call and the cost for the replacement wire. A lot of times they would want regular wire again.
Long story short, it saved me loads of work to eat the cost. It however did not help me financially.

The business here was/is so competitive that you often had to go bare bones and that left no room for an extra $20 or so to cover things like this. Compared to some of the other guys I worked around, I often had one of the lowest go back rates, due to things like that.
Sorry, I got off topic I still like talking about the good old days. I could go on and on.

ps. Cat 5 is slightly cheaper than Cat6. I would choose 6 over 5. I have found as I get older I much prefer working with 6 vs 5, its easier to see and work with with big old hands.

 

[JDB]

Yea I’ll definitely go proper exterior grade cat6, no point quibbling over a few dollars/pounds to do it right!

I’ll likely bite the bullet and run cable between all 4 rooms with kit and do them all exterior. As it happens all the routers are on an exterior wall so easy to run a cable round the outside of the house and not mess about with interior work to hide it!


Regards reboots, what is interesting (and annoying) is if the Node is power cycled either on its own or along side the AiMesh router (such as power cut) it doesn’t come back up properly and seems to free-wheel on its own selecting its own channels etc. If you delay rebooting he AiMesh router even by a few seconds so it comes up after the Node then all is well!!

This is the script I run via cru @4am on my Merlin router, so it remotely reboots the 2 nodes and then the AiMesh router 5 seconds later.

ssh -y -i /jffs/configs/id_rsa admin@jdb-node1 /sbin/reboot
ssh -y -i /jffs/configs/id_rsa admin@jdb-node2 /sbin/reboot
sleep 5
ssh -y -i /jffs/configs/id_rsa admin@jdb-ap /sbin/reboot

The SSH public key of the Merlin router is installed on the AiMesh router via the GUI and so auto-propagated to the nodes to allow password-less login on all.

I’m going to try disabling it though as so far so good, and maybe instead have a power cut detecting script (maybe using the WAN Up jffs script) to remotely reboot the nodes ONLY if a power blip occurs so forcing them to come up Nodes first. It seems insane to me that a Node reboot breaks it, hopefully these are all teething issues with AiMesh (or maybe even a function of having powerline backhaul?).


Sent from my iPhone using Tapatalk

 

[HuskyHerder]

@JDB

I have the reboot running, along with logging it to a file for my own info.

What I find strange is when the system comes back up, is its reaction to my devices. It may just be normal interaction.

So here's what happens.

The script triggers the reboot from the Merlin Router to the Mesh Router, it occurs on nodes 1-5 with a quick log time stamp between each node. I enabled the logging due to one node not rebooting when I manually triggered the script during its creation (all good now). Just like yours, a final delay of 5 seconds before the primary AiMesh Router reboots.

When it comes back up the WiFi is very unstable, more so than after a normal reboot. Though I have never took the entire system down all at once!
By unstable when the wifi restarts, my Mac attempts to connect, succeeds and then disconnects, it repeats this 5 times in a row before things stabilize and I can maintain a connection. Even if connected directly to the AiMesh Router and not a node. Kinda funny it does this about 5 times and I have 5 nodes ???

Just wondering if you have seen anything like that.

Spoiler: my script log

Merlin_scripts.log
12-24-2018 13:30:04 Merlin Router - start up
12-24-2018 13:30:05 Scheduled - cru jobs
12-24-2018 13:32:05 Pushover - sent
12-24-2018 13:35:00 AiMesh - reboot start
12-24-2018 13:35:01 node1
12-24-2018 13:35:02 node2
12-24-2018 13:35:03 node3
12-24-2018 13:35:03 node4
12-24-2018 13:35:04 node5
12-24-2018 13:35:10 AiMesh Router
12-24-2018 13:35:10 AiMesh reboot - complete

 

[JDB]

Yea I think it’s normal for it to have a wobbly few mins after a full reboot. It seems to be as it registers each node on the mesh and re-checks for the best channel to use.

I’ve run Wi-fi scans during a reboot and you see the nodes come up on various channels before they all start to align channels once the AiMesh router takes control.

I guess it must bounce the mesh each time it brings a Node onboard and/or changes channel.


Sent from my iPhone using Tapatalk