On 2017-03-15 03:32, David C. Rankin wrote:
On 03/14/2017 04:43 PM, Carlos E. R. wrote:
It seems that CPU overheats, and the kernel throttles the CPU down. But within a second it says that the temperature is back to normal, and throttling is disabled. Makes no sense.
Somewhere it also says that it is preparing an email.
It trips the temp limit, sends an e-mail and then subsequent to that checks again and is below the limit, e.g.
Yes, but: CPU 6 too hot at 2017-03-14T11:11:33.161797 CPU 6 normal at 2017-03-14T11:11:33.169789 8 milliseconds later! That's impossible!
If this was the first I've seen of the error, I would be looking at cleaning dust bunnies out of the fan screens and making sure there wasn't a rat's nest around the CPU....
Yes, me too, but the time difference points at bugs. One way to have both events so near is noise around the trip point. No hysteresis cycle designed in the detection. It would cause a storm of events, and in fact, there is such: 2017-03-14T11:11:33.429398-07:00 marcslaptop mcelog[816]: mcelog: Too many trigger children running already -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))