Flint 4 - Main Mesh with two nodes …
Flint 3 - Node Ethernet
Flint 3 - Node Ethernet
Also same issue on WiFi. Crash every day or two.
This log actually catches the failure quite clearly.
At about 08:57:25, the Flint 4’s MediaTek Ethernet subsystem starts locking up:
NETDEV WATCHDOG ... eth2: transmit queue 2 timed out
Then it repeats continuously on both eth2 and eth1, with the timeout climbing from ~5 seconds to ~65 seconds. Pasted markdown Pasted markdown
What happened
Immediately before that, your mesh/backhaul topology was changing. At 08:56:36 the Flint created a WDS/mesh interface (rai4.sta4), added it to the LAN bridge, and eventually put it into forwarding state. Pasted markdown
Then you start seeing repeated topology-change BPDUs. Shortly afterward, the physical Ethernet link itself drops and comes back:
08:57:30: LAN1 link down
08:57:33: LAN1 comes back at 2.5 Gbps
08:57:35: both Flint 3 nodes obtain DHCP addresses again. Pasted markdown Pasted markdown
But the Ethernet transmit queues don't recover. The watchdog errors continue over and over. That's why the network appears to effectively freeze/crash even though the router itself hasn't necessarily rebooted.
The interesting part
This looks much more like a Flint 4 beta firmware / MediaTek Ethernet-driver + mesh/backhaul problem than an ISP problem.
In particular, you've now got wired Ethernet backhaul between the Flint 4 and Flint 3s, while the firmware is also creating WDS/mesh bridge interfaces. The repeated:
received tcn bpdu
topology change detected
is consistent with the bridge repeatedly reacting to topology changes. Pasted markdown
Then the MediaTek Ethernet driver goes into the fatal-looking condition:
NETDEV WATCHDOG: transmit queue ... timed out
That's the smoking gun.