RT68U/ 384.8_2 : radios, openVPN turning themselves off

Update: issue corrected with ~1 wk of uptime. Fixed by either new 384.9, factory default reset, or setting radio channel control to manual or a combination of the above.
Thanks for the assistance.

This problem has recurred, with failures becoming much more frequent, despite default resets and an update to 384.10_2 (with reset). Normally, with a change of seasons, I'd suspect an overheating issue, and cpu temp is currently reading 82C, despite cpu utilization in the single digits. I'm running no add-ons or QoS, router level VPN or Aicloud, only DNS filtering.

another piece of the log as it happens:

Apr 16 21:44:36 rc_service: watchdog 236:notify_rc stop_obd
Apr 16 21:44:37 kernel: br0: port 2(eth1) entering forwarding state
Apr 16 21:44:37 kernel: device eth1 left promiscuous mode
Apr 16 21:44:37 kernel: br0: port 2(eth1) entering disabled state
Apr 16 21:44:37 kernel: br0: port 3(eth2) entering forwarding state
Apr 16 21:44:37 kernel: device eth2 left promiscuous mode
Apr 16 21:44:37 kernel: br0: port 3(eth2) entering disabled state
Apr 16 21:44:40 kernel: device eth1 entered promiscuous mode
Apr 16 21:44:40 kernel: br0: topology change detected, propagating
Apr 16 21:44:40 kernel: br0: port 2(eth1) entering forwarding state
Apr 16 21:44:40 kernel: br0: port 2(eth1) entering forwarding state
Apr 16 21:44:40 kernel: device eth2 entered promiscuous mode
Apr 16 21:44:40 kernel: br0: topology change detected, propagating
Apr 16 21:44:40 kernel: br0: port 3(eth2) entering forwarding state
Apr 16 21:44:40 kernel: br0: port 3(eth2) entering forwarding state

Suggestions?

typically an RT-AC68U has about 70-75°C (latter with 1200MHz CPU clock), more than 80 are the reason for your failures in my opinion.
Often there is a gap between CPU and the large heat sink.
If you can correct this it may work propper.

There are good threads and videos or pictures about this issue, you just have to search for them.

Grisu said:

typically an RT-AC68U has about 70-75°C (latter with 1200MHz CPU clock), more than 80 are the reason for your failures in my opinion.
Often there is a gap between CPU and the large heat sink.
If you can correct this it may work propper.

There are good threads and videos or pictures about this issue, you just have to search for them.

Click to expand...

Have never seen those low temps on an RT-AC68U and I've installed many. Closer to the 83C to the 86C range. Not have had many failures either. Actually, just one. And that one was used right beside a fireplace that was used 'a lot'.

I had 5 different RT-AC68U in the past and 2 DSL-AC68U, and they all have been in this range after I corrected the cooling problem on 2 units.
And my 2 RT-AC86U had been far below 70°C out of the box, despite the much more powerful processor.

Faulty power adapter can be an issue too!

Grisu said:

I had 5 different RT-AC68U in the past and 2 DSL-AC68U, and they all have been in this range after I corrected the cooling problem on 2 units.
And my 2 RT-AC86U had been far below 70°C out of the box, despite the much more powerful processor.

Faulty power adapter can be an issue too!

Click to expand...

Your ambient temperatures are much below what they are in my area then.

Agree with the RT-AC86U's. They are more powerful and run a lot cooler too. In my experience in the high 70's and low 80'S C though.

I've never seen a faulty power adaptor make a CPU run hotter though? At least not on a continuous basis, for a one-shot and destroy the CPU, motherboard and/or other circuits as an example, yes. But not one that works for years on end.

I will take the unit apart and check the sinks.
Any clues from the logs?
RMerlin mentioned AiMesh, which I don't have settings for AFAIK and haven't enabled.

Often times a mysterious loss of settings can be caused by low nvram space. You don't have a lot of features, but do you for instance have a lot of DHCP static assignments? What's the nvram utilization on the Tools page?

I've got it apart now so I can't give you an answer on the nvram, but I did have almost every device set as bound IP/MAC in network map.

Yes, one pad had ~ 50% contact with the sink. the pcb appears bowed concave in the direction of the chipsets.

With some careful bending I was able to correct the contact problem, but unknown if this is a source of my issues at all. CPU temp down to 80C, radios at 50C, without active cooling. Laser shows surface HS temps at around 40C.

to john9527:
NVRam 51162 / 65536 bytes, but after a recent factory default reset
JFFS 1.75 / 62.75 MB

ram: 195mb free, 1.1mb buffer, 7.81 cache, no swap

UserUnknown said:

With some careful bending I was able to correct the contact problem

Click to expand...

I did the same on mine , radios ok with 50 and 55°C. You can place some thermal paste additional to the pads, should go below 75 then.

L&LD said:

I've never seen a faulty power adaptor make a CPU run hotter though?

Click to expand...

And power adapter was meant to be responsible for reboots, not high temps!

UserUnknown said:

to john9527:
NVRam 51162 / 65536 bytes, but after a recent factory default reset

Click to expand...

That looks fine right now, but you may want to double check again incase of another fail.

One other thought....do you have any clients running Android 9 or 8.1? These have been reported to cause a restart of wireless when they connect.