DL380 G7 Uncorrectable Machine Check Exception

[Nicolai Rasmussen]

We have a bunch of new DL380 G7, and so far we've seen this happen on 3 of them:

Uncorrectable Machine Check Exception (Board 0, Processor 1, APIC ID 0x00000001, Bank 0x00000005, Status 0xF2000000?, Address 0x00000000?, Misc 0x00000000?)

Running Windows Server 2008 R2. This error causes the machine to reboot. The server is idling, as it has not yet been put into production. We have the Hyper-V role installed, but no VMs yet.

[Erdogan Temur]

Hi,


RESOLUTION
Ensure that only one NIC port is enabled for PXE boot.

Since the default System ROM setting has only one port enabled for PXE boot, perform the following if additional ports were enabled:

Reboot the server. During POST, press F9 to enter RBSU.
Select "System Options."
Select "Embedded NICs" and select only one NIC for "Network Boot." Ensure that the other NICs are set to "disabled" (that this does not disable the device; it only removes it from being available to PXE boot).


That will also fix the problem :) Remember to assign points to answers that helped in your problem. So the forum stays alive...

[Nicolai Rasmussen]

I appreciate your suggestion, but I have already read the advisory that you are referring to myself. Perhaps I was a bit vague in my description of the problem, but the PXE boot issue is only an issue IF/WHEN you try to PXE boot, and multiple NICS are enabled for PXE boot. I'm not trying to PXE boot, nor have I changed the default setting (only 1 nic enabled for PXE boot). - I'm simply trying to keep my servers from spontaniously rebooting :)

If others find this (un)helpful, here's the advisory that I'm referring to:

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSeriesId=466340&prodTypeId=12169&objectID=c02251106

[Nicolai Rasmussen]

Sorry, wrong link :P

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c02039369〈=en&cc=us&taskId=135&prodSeriesId=3794183&prodTypeId=329290

[Jan Soska]

Hello Nicolai - is there any difference between setup of 3 "bad" and other dl380g7's ?

jan

[Nicolai Rasmussen]

Hi Jan,

no they are all setup using the same procedures. Same OS and same bios settings.
We don't change the default bios settings, since all virtualization features is enabled per default. All servers have been upgraded to the latest bios version and none of them have any PCI cards installed.

[Jan Soska]

Hmm, could you more for testing purposes system drives from 1 bad to 1 good? If good one becomes bad - there is definetelly problem in your system OS config as only drives very changed... If issue stays on bad one, there is hw problem and contact HP?

Jan

[Nicolai Rasmussen]

We've replaced the motherboards on the faulty servers and we have not seen the errors since.

[Nicolai Rasmussen]

It turns out that this WAS in fact a known issue (thanks for NOT sharing that HP!!!).
The latest Bios fixes this issue:

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=dk&prodTypeId=15351&prodSeriesId=4091412&swItem=MTX-7f6a47f7f59845989479d4dc2e&prodNameId=4091432&swEnvOID=4064&swLang=13&taskId=135&mode=4&idx=2

[Nicolai Rasmussen]

Firmware 2010.08.16 fix this issue. It was - at we at some point expected - CPU related. The latest microcode from Intel fixed it.

- It appears that this is NOT the finale fix for this issue. We still have servers with UMCE errors after the bios upgrade. Two servers had the motherboard replaced, and it hasn't happened on them since, so I would go with that solution for now...