DL160 G6 can't read the CPU temperature

[Niki17]

Hello all,

I will really appreciate some help on this.

I have a DL160 G6 server. It runs with the latest firmware available at this moment, latest iLO firmware, latest P410 firmware.

I decided to upgrade the cpu's from 2xE5620 to 2xL5640. The upgrade went fine, everything is working as expected until I noticed in ESXi 5 health status that both cpu's are showing exactly 30 degrees and this value never changes, ever.

I then checked the health status in the bios, same story, always 30 degrees and the value never changes.

Then I asked myself, it's maybe different sensor in L5640, maybe the Motherboard/chipset/bios can't read the data.

I reverted for test to the old E5620 and it was exactly the same story...30 degrees per cpu and the value never changes.

Does anyone have any idea what's the issue here and any advise how this can be solved? Maybe someone with the same issue reverted to old version of the bios which fixed the issue or something like this?

Best Regards,

[TVVJ]

Hello,

Was the server's BIOS updated?

Regards,

Vijayasarathy

[Niki17]

Hello,

yes the bios was updated, but I revert it just to be sure it's not the reason, I upgraded the ILO firmware aswell, tried every possible combination...it always shows 30 degrees. But if I connec to the ILO, during the bootm the first 15-20 seconds while there is still nothing on the monitor, the monitoring sensors menu of the ILO shows the correct temperatures of the CPUs, then something happens internally, and these temperatures going to 30 degrees and never changing again. At the moment I'm with the latest bios firmware and ILO firmware, but I tried with older bios from 2012...same story.

It seems to be some really nasty bug somewhere and I wonder how someone else didn't discover and reported it yet. or maybe someone did but with no results.

[Niki17]

Further investigating this issue,trying with different versions of the ILO firmware, I noticed something very interesting: when I boot in DOS, and run the rompaq.exe and it shows the "Ready to flash!" screen

if I access the ILO, the CPU temperatures are read correctly, they are changing depending on the ambient temperature etc, everything works as expected.

I can leave it on this screen for hours, and everything will work just fine monitored through the ILO.

Once some ILO firmware is installed, and server boots in DOS or some other OS, the problem is there.

The question, in what mode this rompaq puts the server so the cpu temperatures can be read correctly?

With these findings it should be crystal clear for someone knowing the ILO code what's going on. 

[flux103]

I am having this exact same issu with dual E5520's cpu temps get over 80C sometimes!