temperature concerns
I would like to find out what is considered "safe operating" temperatures for components in my V880 servers. I have read all the documentation on critical temperatures and at what temperature the component will be shut down, but I cannot find anything that indicates at what temperature I should be concerned. Currently I have a situation where my CPU's are running between 140 - 167 F, the "critical level" is 185 F. Should I be concerned that these CPU's are close to the critical level or is this normal?
Also I have a 3800 that is showing 77 Celcius on the SB0 SDC, and other components on SB0 are showing between 59 - 70 Celcius, this seems very hot to me, but the system is saying everything is "ok".
Any help or pointers to documentation would be appreciated.
Angela
[808 byte] By [
aguthrie] at [2007-11-25 23:39:23]
# 1
Hi Angela,
If the CPUs are too close to the "Critical Temperature" - you should be worried. There are several reasons why the temperatures might high.
There could be environmental factors that might be causing the problem. For instance, your datacenter chillers might not be running properly, and the temperature might be rising gradually.
The air-flow through server could be interrupted and thus impacting cooling. I've had a server crash on me because someone stuck a piece of paper up front (I'm talking about a V440) where the air-flow inlets are.
You might also want to check temperatures on a few other (comparable) servers in your datacenter. Are they also running at identical temperatures?
If you want to find the thresholds of your servers (in many cases prtdiag doesn't show it) -- you might want to investigate the command "prtpicl" (this queries the "picld" daemon -- this is available only on the SunFire series iirc) . I don't remember the exact command, but prtpicl -c <class> will give you indepth info about that hardware class.
prtpicl -c cpu -v (will give you details about the cpu class from picld) and should contain the two extreme temperatures of the cpu.
HTH
