Service Guide

BIOS Setup Utilities

130 Intel

Server System P4000IP and Intel

Workstation System P4000CR Service Guide

Option Values: 20

All

None

Help Text:

Threshold value for logging Correctable Errors (CE) – Threshold of 10 (default) logs

10th CE, "All" logs every CE and “None”’ means no CE logging. All and None are not

valid with Rank Sparing.

Comments: Specifies how many Correctable Errors must occur before triggering the logging

of a SEL Correctable Error Event. Only the first threshold crossing is logged, unless “All” is

selected. “All” causes every CE that occurs to be logged. “None” suppresses CE logging

completely.

When Rank Sparing RAS Mode is configured, “All” and “None” are not valid, so they will not be

presented as choices.

This threshold is applied on a per-rank basis. The Correctable Error occurrences are counted

for each memory rank. When any one rank accumulates a CE count equal to the CE

Threshold, then a single CE SEL Event is logged, and all further CE logging is suppressed.

Note that the CE counts are subject to a “leaky bucket” mechanism that reduces the count

as a function of time, to keep from accumulating counts unnecessarily over the term of a

long operational run.

This is also the Correctable Error threshold used when Rank Sparing RAS Mode is configured.

When a CE threshold crossing occurs in Rank Sparing Mode on a channel which is in

Redundant state, it causes a Sparing Fail Over (SFO) event to occur. That threshold crossing

will also be logged as a Correctable Error event if it is the first to occur on the system.

An SFO event causes the rank with the error to be replaced by the spare rank for that

channel, and the channel goes to a non-redundant state (with a “Redundancy Degraded” SEL

Event logged). There may be an SFO for each channel in the system, although only the first

one can be logged as a CE event.

9. Memory Power Optimization