Upgraded Flint2 to version 4.9 last week (standard version, not OP24). Today, total connection loss. Even after two reboots.
Router details (v4.9 firmware):
Hostname: GL-MT6000
Model: GL.iNet GL-MT6000
Architecture: ARMv8 Processor rev 4
OpenWrt Version: OpenWrt 21.02-SNAPSHOT
Kernel Version: 5.4.238
Key configs:
Hardware acceleration: OFF
Adguard Home: ON (no fancy config, just the two switches on the front end UI set to ON)
For the rest, I’m using out of the box available glinet admin configs, nothing fancy, no changes to backend advanced settings. Plain 4.9 firmware on OpenWRT v21.
Used Gemini to troubleshoot. Here’s what it says about the logs:
Mediatek driver crash (in a loop, for the last 10 hours whilst I was at work):
Thu Jun 25 20:30:24 2026 kern.err kernel: [35523.904570] mtk_soc_eth 15100000.ethernet eth1: transmit timed out
Thu Jun 25 20:30:24 2026 kern.info kernel: [35524.547105] is_qfsm_hang = 1
Thu Jun 25 20:30:24 2026 kern.info kernel: [35524.552680] is_qfwd_hang = 1Ethernet ports froze completely:
[35337.024570] mtk_soc_eth 15100000.ethernet eth1: transmit timed out
[35341.120564] mtk_soc_eth 15100000.ethernet eth0: transmit timed out
[35341.091099] is_qfsm_hang = 1As soon as that structural timing fails, the physical network driver locks up. Your logs show both eth1 (WAN/Internet link) and eth0 (LAN/Local link) throwing "transmit timed out" errors.
The line
is_qfsm_hang = 1indicates that the processor's Queue Forwarding State Machine has structurally hung. At this exact timestamp, the router's network ports are physically broken and cannot process data.Local system queue (
dnsmasq) instantly caps out at its maximum limit of 150 parallel connections, creating a dead-end network traffic jam:Thu Jun 25 20:30:39 2026 daemon.warn dnsmasq[23864]: Maximum number of concurrent DNS queries reached (max: 150)
What’s going on?
Not a heat issue - I’ve had two fans sandwiching this device since unboxing last year.
I have had frequent (weekly / bi-weekly) internet or wifi disconnects ever since receiving. As such, I setup a weekly reboot schedule. Disconnects still infrequently occur and usually last 1-3 minutes. Today, total 12 hour blackout whilst I was at work. Not an ISP issue as the modem was up and running all day.
Got home, switched off AdGuard, everything went back to normal. Switched on AdGuard, everything still working.
All I want is stability. Frustratingly I don’t seem to be able to achieve stability with this router.
Gemini telling me to switch to OP24 branch. Or for even more stability, to just overwrite the full glinet firmware with the latest stable stock OpenWRT 25.
What am I missing? Is this poor 4.9 coding? Should I switch to OP24 (some reddit forums seem to indicate similar issues with adguard in OP24)? Or should I bite the bullet and overwrite glinet’s firmware with a more stable OpenWRT community code (but loose out on all the nice simple interfaces)?
Advice sought.
