These forums are locked and archived, but all topics have been migrated to the new forum. You can search for this topic on the new forum: Search for CPU temp threshold warning on the new forum.
Does anyone know how to adjust the threshold temp warning in CentOS 4.4? I have some new 1Us that are in a constant state of warning and I can't even see what the threshold is set at. (yes, it still continues with the lid removed.) I'd be suspect if it weren't for all of them doing it.
Hey Dan,
It's a BIOS thing, but by the time it's warning at the Linux kernel level (I suspect), it actually is at pretty dangerously high levels.
The Linux kernel, I believe, only begins issuing messages about temperature when the CPU itself downclocks to protect itself from damage (which is not adjustable, as far as I know).
Notice there's all these "as far as I know" and "I believe" wishy-washy language in my comment...I don't know this stuff for certain. But I do know that when I paid a lot more attention to hardware issues a couple of years ago, it was like pulling teeth to find out <i>anything</i> about temperature under Linux, except true heat problems where the CPU was lowering frequency in order to avoid damage. lm_sensors had to be installed and configured to get anything useful out of the system--and it only ever worked on about 10% of the systems I've had in my grubby hands.
So, I'm pretty sure you really do have a temperature problem with those boxes, despite it being common across several.
You didn't post the actual message, so I can't be sure of this--but I'd know the CPU throttling message if I saw it.
Oh, yeah, all of that said, check for a BIOS upgrades with the motherboard manufacturer. Sometimes they do something stupid that is incompatible with newer CPUs and they fix it in a flash update.
--
Check out the forum guidelines!
yup, It's running at a modulated CPU level
I'll have to see about rerouting the wiring harness as the bundle sits in front of three fans.
Thanks, I was hoping I wasn't losing it.
Dan
are these dell servers? if so install OMSA from linux.dell.com. That will allow you to log into the machine and read the temperatures and also setup alerts that are sent via smtp. However if your cpu's are throttling you have some serious issues. either your cooling is insufficient or the fans/heatsinks are very dirty. It's time to pull those machines and open them up and see what is going on.
I bought these as a bulk purchase and it turns out that the heatsinks are horizontal and there's only about 3/16" clearence to the lid from the fan. IE nowhere for the air to go. So the option is either the old cardboard tunnel to the CPU or vent the lid above the CPU. I thought those days were past.