Weird fan behaviour on a 2950
dirkjan at ruby-sparks.nl
Fri Jun 8 04:03:35 CDT 2007
Tony van Lingen wrote:
Dutch or from recent Dutch heritage? Because my name is without the
second capital and only people with some sort of Dutch background would
probably type it this way ;).
> I don't think its purely cosmetic. Raghavendra says:
Well, the way I read it, they don't have absolute sensors anymore, but
relative ones. I've looked at the current values and they are very
reasonable (so the negative temperature shows a realistic temperature
given the overheating point and the temperatures in the data center).
Also if there is no CPU present in a socket, it shows a totally
different value (always 50 as the temperature, tested on various
machines), indicating it does read something from the CPU.
I've also find no reason behind the fact that the bit is set. It could
be only because it is confusing, or because the data actually doesn't
make any sense (in which case it would be very coincidental that the
temperatures actually look like reasonable values). Also my primary
concern is not even the temperature, but far more the "ok" status. If
that is readable, I would like to be able to read it without for example
patching more recent versions of ipmitool.
I think the only one here who could shed some light on this would be
More information about the Linux-PowerEdge