[CentOS] RAM errors after kernel-update

Tue Oct 6 13:28:10 UTC 2009
Michael Schumacher <michael.schumacher at pamas.de>

Hi,

I updated a server yesterday from

"kernel 2.6.18-128.7.1.el5xen"  to "kernel 2.6.18-164.el5xen"

After rebooting, my message log is flooded every second or so with this error messages:

Oct  6 14:52:20 xenserver1 kernel: EDAC MC0: UE row 0, channel-a= 0 channel-b= 1 labels "-": NON-FATAL recoverable (Branch=0 DRAM-Bank=0 Buffer ID = 0 RDWR=Read RAS=0 CAS=0 NON-FATAL recovera
ble Err=0x2000 (FB-DIMM Configuration Write error on first attempt))

and

Oct  6 15:17:23 xenserver1 kernel: EDAC MC0: CE row 0, channel 0, label "": Corrected error (Branch=0 DRAM-Bank=0 RDWR=Read RAS=0 CAS=0, CE Err=0x10000 (Correctable Non-Mirrored Demand Data E
CC))

The machine is a new Tyan S5397 mobo with 16GB Kingston RAM KVR667D2D4F5K2/8G

Removing and replacing memory to different locations doesn't make any
difference.

After some digging, I noticed that the new kernel has added support
for the i5400 chipset. I found some reference that the new kernel has
this error reporting capability the old one hadn't.

Question1: how many recoverable RAM errors are acceptable?
Question2: The error appears always with the same id in the error
message. Mobo problem?
Question3: Are there any recommended BIOS settings to operate the RAM
slower to see if the problem disappears?
Question4: Any other proposals.

Being located in Germany makes the "just return it to the dealer"
proposal quite unattractive.


best regards
---
Michael Schumacher
PAMAS Partikelmess- und Analysesysteme GmbH
Dieselstr.10, D-71277 Rutesheim
Tel +49-7152-99630
Fax +49-7152-996333
Geschäftsführer: Gerhard Schreck
Handelsregister B Stuttgart HRB 252024