Story Time! Made the jump to 3006 - Everything good until...

[CaptnDanLKW]

I feel like every time I post something, its a looong story.

Previous config - AX86UPro - 388.8_4 AC86U - 386.14_2. Stable for year(s) (before AX86U Pro, I was 2x AC86U in the identical setup)

Environment:

• I live in a very low-density area
• 5-6 other networks seen, all on the fringe of range
• Optimal placement of my router and nodes - large property
• Main Router north side of house, Garage Node on south side and Node on East side in an Outbuilding
• Nodes are ethernet backhaul

Made Switch to 3006.102.5

• Complete config wipe and manual reconfig of everything
• Decided to stay with the 192.168.50.x scheme - previously I'd always go back to 192.168.1.x
• Custom Guest Network Pro - 2.4GHz only VLAN 52 created and added about ~20 IoT devices with DHCP reservations
• All Devices connected and bounced around until they stabilized - (on the closest node as they should)
• SmartConnect and Mesh enabled. (SC works for me for at least 4 years without issue. I have tuned my setting to my specific environment)
• Ran without addons for about a week
• Installed my addons, including scribe and my very specific logging preferences - including nodes sending their messages to Scribe/syslog-ng
• Been running good for about a month; devices all connected to the desired noded
  • wi-fi connection times were all lengthy and consistent. - meaning once they initially connected, they never bounced or dropped

Out of the blue:

• Couple days ago, everything went haywire
• Discovered when my garage doors showed offline
• Then realized most clients were not and would not connect to either of the nodes (fixed location, strongest signal)
• One or two clients would say and show as connected to the Nodes in the UI never checked if they actually worked
• My logging showed that the devices kept auth + Assoc successfully to the nodes but would drop immediately. Thousands of these from every mac address that wouldn't connect

Sep 29 10:48:48 RT-AC86U-DC38 wlceventd: wlceventd_proc_event(511): wl0.2: Disassoc CC:8C:BF:xx:yy:zz, status: 0, reason: Disassociated because sending station is leaving (or has lef
Sep 29 10:48:48 RT-AC86U-DC38 wlceventd: wlceventd_proc_event(530): wl0.2: Auth CC:8C:BF:xx:yy:zz, status: Successful (0), rssi:0
Sep 29 10:48:48 RT-AC86U-DC38 wlceventd: wlceventd_proc_event(559): wl0.2: Assoc CC:8C:BF:xx:yy:zz, status: Successful (0), rssi:0

Troubleshooting:

• Rebooting the nodes didn't help.
• Rebooting everything didn't help.
• Went down the Rabbit hole changing professional settings, toggling AX mode, toggling auto/legacy WIFI mode, toggling OFDMA/MU-MIMO
• I removed one node and added it back, no change
• Removed the other node and added it back, no change
• Repeated above steps over and over, doing the same thing for hours expecting a different result

Dumbfounded, I removed both nodes and shut them off and rebooted the main router. Tested to see how many devices would successfully connect. Everything that could connect did so and worked (even if the signals were poor-very poor)

I let everything settle down for a few hours, added the nodes back and everyone was happy again, like nothing ever happened. AIMesh eventually put everyone back on the node with the strongest signal and things have been good for a few days now.

Very odd.

The only message logged on the nodes that could potentially be an indicator of the underlying issue were these (seemed to be when I added or removed or changed some setting on the Wi-Fi Professional Section) (lots of rabbit hole troubleshooting):

Sep 29 10:46:39 RT-AC86U-DC38 kernel: wl1.1: promiscuity touches roof, set promiscuity failed. promiscuity feature of device might be broken.
Sep 29 10:46:39 RT-AC86U-DC38 kernel: wl0.1: promiscuity touches roof, set promiscuity failed. promiscuity feature of device might be broken.

I can only hypothesize that something went really wrong with the AI Mesh code that pushes the settings down to the nodes and it took these specific steps to clear:
Remove all nodes
AND reboot the main router
THEN add them back (since removing them and adding them back without rebooting the main router didn't work).

So that's the story - Probably too nuanced for general Q&A but maybe the TL;DR is if you moved to 3006 Codebase and still have legacy devices, and your legacy mesh nodes freak out and nothing can connect to them, you may need to remove them, reboot your router (important step) and add them back.

Oh, and I'm happy that Guest Network Pro's SDNs (At least the Wi-Fi ones) are propagated to the 86U legacy devices just fine. This was my biggest worry and why I waited so long to upgrade.

Thanks for reading, hope you enjoyed story time.

 

[OzarkEdge]

If it comes back, assume its one of the AC86Us dying and turn that one OFF to test running without it.

OE

 

[Ronald Schwerer]

I had pretty much the same issue after installing 3006.102.5 on my RT-AX88U Pro with around 60 devices and 4 nodes. My meshed devices got setup (automatically) by the 3006.102.5 installation to be on IoT VLAN 52. It initially ran very good for a few days then I noticed devices falling of continuously. Tried lots of things but ended-up making things worse. After fighting for 2 days to at least get my 10 security cameras working, I tried moving a those devices over to my main SSID from the Guest on on VLAN 52. It was a pain individually resetting the devices since some were hard to reach. Well every device I reconfigured worked flawlessly. So. I continued reconfiguring most of the other devices to the main SSID. I know I've opened myself up for a hack attack, since there's no isolation between the trusted and untrusted devices, but since I was coming from the old 3004.xxx AiMesh (where isolation was just a joke), I don't think I'm less secure than that.