IPMI temperature sensors

Chris Cooke cc at inf.ed.ac.uk
Mon Feb 1 04:53:10 CST 2010


I have a question about IPMI temperature monitoring, and I'm hoping
that someone on the list can help.  We have various Dell Poweredge
servers which we want to shut down cleanly when the temperature gets
too high. The idea is to do a clean shutdown before any damage is done
and before they would reach the point at which the systems would start
cutting off their own power to preserve the hardware. The servers
mostly run Scientific Linux 5.

What's the appropriate temperature at which to shut down the machine?
I'm guessing that the ambient temperature would be the thing to look
at.  When I query the ambient temperature sensor on an R710 with
'/usr/sbin/ipmi-sensors -v -s 5' I see something like this:

Record ID: 5
Sensor Name: Ambient Temp
Group Name: Temperature
Sensor Number: 7
Event/Reading Type Code: 1h
Lower Critical Threshold: 3.000000 degrees C
Upper Critical Threshold: 47.000000 degrees C
Lower Non-Critical Threshold: 8.000000 degrees C
Upper Non-Critical Threshold: 42.000000 degrees C
Lower Non-Recoverable Threshold: NA
Upper Non-Recoverable Threshold: NA
Sensor Min. Reading: -128.000000 degrees C
Sensor Max. Reading: 127.000000 degrees C
Normal Min.: 11.000000 degrees C
Normal Max.: 69.000000 degrees C
Nominal reading: 23.000000 degrees C
Sensor Reading: 25.000000 degrees C
Sensor Status: [OK]

What exactly is the Upper Critical Threshold?  Is this the temperature
at which the machine may shut its power off?  Should I be thinking of
starting a clean shutdown when the ambient temperature reaches the
level of the Upper Non-Critical Threshold? And what exactly is the
Upper Non-Critical Threshold anyway?  I've been looking for an
explanation of these terms and haven't yet found one.  I've looked in
various Intel and Dell and IPMI docs but with no success so far,  
unfortunately.

Can anyone point me at sources of help, or have you tried this already?

If you're wondering, we're in a shiny new building and they haven't
entirely ironed out the problems with the server room's cooling system!

	-- Chris Cooke.

Computing Officer, School of Informatics, University of Edinburgh.




-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.



More information about the Linux-PowerEdge mailing list