Table 33. Memory ECC Error Handling — Runtime, Non-Redundant Configuration
Error Scenario System Event Log
(SEL)
DIMM Fault LED1
System Fault LED
IPMI Memory RAS Behavior System Operation
Correctable Errors
CE < Threshold2
CE SEL message DIMM LED:
No change
System Fault LED:
No change
Set DIMM State
Set Memory RAS Redun. State
Set RAS Config Information:
No change
The system continues to operate.
CE = Threshold CE SEL message
CE Threshold
Reached SEL
message
CE Logging Stopped
SEL message
DIMM LED:
On for the failed FBDIMM
only
System Fault LED:
Amber blink: More than one
FBDIMM installed.
Amber on: One FBDIMM
installed.
Set DIMM State:
DIMM failure status = Y
DIMM disabled status = Y
Set Memory RAS Redun. State
Set RAS Config Information:
No change
The system continues to operate normally, but masks all
correctable memory errors.
CE > Threshold No action DIMM Fault LED:
No change
System Fault LED:
No change
Set DIMM State
Set Memory RAS Redun. State
Set Memory RAS Configuration:
No change
Operating system continues to operate normally.
UE UE SEL message
identifying the
FBDIMM location
DIMM Fault LED:
On for the lock stepped pair
or for a single FBDIMM,
depending upon the mode of
operation
System Fault LED:
Amber on
Set DIMM State:
DIMM failure status = Y
DIMM disabled status = Y
Set Memory RAS Redun. State
Set RAS Config Information:
No change
Chipset initially reports recoverable error. After initial
recoverable error is reported, the chipset issues AMB fast
reset and retries the memory transaction. If either the fast
reset or retry fails. The BIOS logs a SEL record for
uncorrectable ECC memory error and halts the system
with an NMI.
See Section 14.2.14.4.3 for more information.
Notes:
1. When an FMDIMM pair is operating in lock-stepped mode and one of the FBDIMMs fails, the BIOS lights the DIMM Fault LED of both FBDIMM modules because the
failure cannot be isolated at the individual FBDIMM level in this mode.
2. The correctable Error logging threshold for non-redundant configurations = Ten Correctable Errors.