Help resolving MCE crash

[NeonOverflow]

Hi.

I'm trying to set up pFSense on my Dell Optiplex 390, but I keep running into MCE crashes. I had used the computer as a pFSense box previously, but I started facing problems after I switched to an Intel ET2 Quad Port NIC. Ever since that switch pFSense, after making it's boot chime, would work, usually for a few minutes but sometimes less than 45 seconds, and then lock up and stop working altogether. Once it stopped working, it would at times kernel panic and completely restart, but other times it would stay booted but not show any signs of life unless you hard reset it.

Below is a snippet of the error that was logged, you can find an identical error by searching for it. Most occurrences you'll find will have happened on the exact same computer (Dell Optiplex 390) by users who also happen to be trying to use a quad-port Intel NIC. I was also able to find a post by a user who had an identical set of problems but was using a 790 series Optiplex instead. This leads me to believe it's a common issue on all Dell Optiplex x90s, but I have no clue how to go about resolving it. I'm aware my NIC is likely disagreeing with my PC for some strange reason, but I'd rather keep both and leave spending more money as a last resort. Any help resolving this would be greatly appreciated. Thanks!

MCA: Misc 0x3ffff
MCA: Address 0x3fff80609d46
MCA: CPU 0 UNCOR PCC OVER internal timer error
MCA: Vendor "GenuineIntel", ID 0x206a7, APIC ID 0
MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000004

[stephenw10]

MCA errors like that are almost always a hardware issue that we can do nothing about in pfSense.
Occasionally you might find some new driver after an upgrade has exposed a hardware issue that wasn't previously being triggered. But even then the only thing you could do is attempt to disable it.

Steve

[NeonOverflow]

@stephenw10: I was aware this wasn't a pFSense issue, I was just hoping someone else had figured out how to solve it since it seemed to be a common issue. Thanks for your response though.

Luckily I did find a solution a few minutes ago after a lot of searching. Apparently it's an SMBus issue on 2nd generation Sandy Bridge and some AMD motherboards from Dell. The solution is to cover pins 5 & 6 (SMCLK and SMDAT respectively) on the front of the card with masking tape. This thread from ServerBuilds has more info. Hopefully that information is useful to anybody finding this thread down the road.

[stephenw10]

Ah, yes I've seen a few users hit that with different things. Nice catch!