DL380 G3 NMI parity error

[Tim.Jackson]

Have a DL380 which recieves the following blue screen of death at random.

*** Hardware malfunction ***
Call your hardware vendor for support
NMI: Parity check/ memory parity error
*** The system has halted***

I have replaced the 4gig of ram with brand new ECC registered chips and still had no luck. The smart array 5i raid card has the lastest firmware on it 2.76 along with the latest driver 5.80.2.32 running 2003 server.

What next steps can be taken in this battle?
Thanks

[KarloChacon · Accepted solution]

hi

does the server have the latest System ROM as well?

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=us&prodNameId=3288130&prodTypeId=15351&prodSeriesId=316529&swLang=8&taskId=135&swEnvOID=1005#12212

also I recommend to update the storport driver

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c01177414〈=en&cc=us&taskId=101&prodSeriesId=428936&prodTypeId=15351

bye

[Tim.Jackson]

Thanks Karlo,

It already had the latest System ROM installed, I have applied the storport driver update, we'll see how this goes.

[GeorgeBell]

This may be a complete red herring, but one of our two DL380 G3s actually failed completely with this error. (Thankfully it wasn't random.) HP's diagnostics failed to come up with an expalantion or cause!!
Having drawn blanks all over, we installed Linux Ubuntu which gave us a detailed installation log indicating that the on-board gigabit network ports were suspect.
We disabled the gigabit ports on the DL380, installed a conventional PCI Network card, and it's been running fine for almost a year now.

[LaneB]

I had this issue very recently in a DL380 G3, while trying to install Windows Server 2003.

The issue seems to be related to adding, in my case, a Compaq NetFlex gigabit fiber NIC. I had originally installed the NIC in the card cage before I started OS installation, and that's when I got the NMI/Parity blue screen.

When I removed the NIC, and simply let the system set itself up with the onboard copper ports, I had no problems at all.

I then shut down, added the fiber NIC back in, and re-booted. Bang! The error was back.

After muttering several impolite phrases, I tried (on a hunch) moving the NIC from the middle slot in the riser cage to the top one. Poof! The problem vanished, and the NIC immediately installed itself on reboot.

I'm no expert at PCI bus design, but my guess is there's a bus-timing issue with some types of NICs. Apparently, the fiber board needs to be in a specific slot in this particular model of server (I've yet to see the issue with any others).

I'll shortly be installing an ML370 G3. I will see if the problem also occurs with it.

Happy tweaking.

[LaneB]

Updating as promised: I finished loading the ML370 G3 last night, and did NOT experience the same (or any other) problems. Installation went flawlessly.

So -- I can only conclude this is something which is found in DL380 G3's.

Happy tweaking.