Machine Check Exception on new Opteron server

humbletech99 · 09-21-2006, 09:41 AM

I set up a new amd64 gentoo server yesterday on an opteron but within a few hours of it being up I got a "Machine Check Exception" and the thing froze up. I had to go to the the local console to see this and then had to hard reboot the machine. It wasn't really doing much at the time other than compiling a couple of things. The server is a dual-cpu dual-core machine (4 cores that is) with 8GB ram and 12 SCSI disks + 2 satas for OS.

The error from the console is below:

Code:

HARDWARE ERROR
CPU 2: Machine Check Exception:                                    4 Bank 4:  f615200133000813
TSC 5ac60e50b6a ADDR 1d251ec00
This is not a software problem!
Run through mcelog --ascii to decode and contact your hardware vendor
Kernel panic - not syncing: Machine check

I have been googling around since yesterday but haven't found anything conclusive

I've tried running mcelog and got the following:

Code:

# mcelog --k8 /dev/mcelog
MCE 0
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a4d0cd72d5a8
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 1
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a56b2eba7649
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 2
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a60591585bda
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 3
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a69ff2a635e8
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 4
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a73a53f42ca9
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 5
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a7d4b6934fdf
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0
MCE 6
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 2 4 northbridge TSC a86f17e0a6a8
ADDR 191b0b000
  Northbridge Chipkill ECC error
  Chipkill ECC syndrome = c12f
       bit46 = corrected ecc error
       bit62 = error overflow (multiple errors)
  bus error 'local node response, request didn't time out
      generic read mem transaction
      memory access, level generic'
STATUS d417c000c1080a13 MCGSTATUS 0
MCE 7
HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
CPU 0 4 northbridge TSC a86f17e0c311
ADDR 23c400000
  Northbridge GART error
       bit61 = error uncorrected
  TLB error 'generic transaction, level generic'
STATUS a40000000005001b MCGSTATUS 0

Does anybody know anything about this?

stress_junkie · 09-21-2006, 10:54 AM

Quote:

generic read mem transaction
memory access, level generic

This makes me think that you had a memory access error. You could try changing memory or you could try running a very stable distro like Debian just to see if you get the same problem. Since you say that you set this machine up yesterday it couldn't possibly be running in a mission critical role yet. I think I'd try using Debian first, then if you get other hardware errors related to memory you could change the memory boards.

If you have no more than four gigabytes of RAM I would use the 32 bit kernel as well.

humbletech99 · 09-21-2006, 10:57 AM

I'm not really gonna install debian just for this, I'd be much more inclined to let it run memtest86 overnight since then I don't have to go through all the compiling and custom settings again.

I was thinking that this is either a CPU problem or a Ram problem. It seems to only happen with Opterons from what i can tell from googling.