We have a IBM P730 machine running AIX 7.1 in a properly air cooled server room. Just recently the fans on the unit kicked into overdrive, they are very loud and spinning at max.
Typically this happens when the server first boots then they normalize.
However for some odd reason, they sound like they are spinning at max RPM.
There is no LED amber light on the control panel or anything in the errpt (though I think it was cleared very recently so if there was something in there - it is now gone).
Is there a way from the OS to control the fan speed? There is no LPARs on AIX, we run the OS on bare metal - no virtualization of any kind.
Since this is a D/R site - I have tried to shutdown and startup the machine (I know seems excessive for a UNIX machine); but it did not resolve the issue.
I also tried the "diag" command in AIX and didn't find any issues.
None that i would know of. As far as i know the fans are controlled by thermo-sensors which kick them up when the system gets (too) hot and slower them once the temperature lowers.
You might try connecting to the service processor port and look at the ASMI. In case the system is not connected to a HMC: you need a web browser on the system you connect with and a crossover network cable to connect your system to it (i use a small netbook with Linux for such purposes). You have to configure an IP address out of the address range the interface is configured with (defaults are 169.254.2.147, 169.254.3.147, 169.254.2.146 and 169.254.3.146, see the IBM documentation.
You will find something in the logs there, even when the orange LED has been cleared - if there was anything, to begin with. To me it sounds like one of these thermo-sensors might be broken/somehow dysfunctional, in this case only the hardware support can help. (You *do* have a hardware support contract, don't you?)
I was thinking the service processor route next, it has been awhile since I have used it - thanks for the notes and link to docs...
We do not have any hardware support on the system, it is scheduled to be replaced in a few months so I just need to squeeze a bit more life out of it.
When I look at the location codes for the P730 here, it shows the sensor should be at "Un -D1"; looks like that's the whole control panel that slides out - i'll try re-connecting that part.
Interesting that I was not able to see in "diag" a way to select and test the thermal sensor, it was not in the list.