IBM Enterprise Server S80, User Manual, Page: 106 / 282

Configure/Deconfigure Processors
The system features Repeat-Gard, an enhanced error detection and prevention
function. This function allows the system to deconfigure processors from the system
configuration if a processor has an unrecoverable error, or repeating recoverable errors.
This prevents the system from using a defective processor.
If a processor is deconfigured, it remains off-line for subsequent reboots until the faulty
processor hardware is replaced. The Repeat-Gard function also provides the option for
users to manually deconfigure a processor, or re-enable a previously deconfigured
processor. See page 71 for a description of the service processor menus used to
configure or deconfigure a processor.
Processor deconfiguration during system boot is a function controlled by the system
firmware which enables removal of processors from the system configuration. If a
processor has a failure which occurs during tests that are performed during the system
boot, or if the processor has had prior unrecoverable or recoverable errors, it will be
removed from the system configuration. The system uses existing hardware Built-in
Self-Tests (BIST) and/or the firmware Power-On Self-Test (POST) to discover and
isolate processor hardware failures during the system boot. The system utilizes the
hardware error detection logic in the processor to capture recoverable and
unrecoverable errors that occurred during prior system operations.
Processor Deconfiguration During Boot
Repeat-Gard utilizes the error detection and isolation capabilities of the processor card
along with the error recording facilities of the service processor to provide a method to
record key fault information. This information is recorded when a processor has reached
a threshold of recoverable errors. It also records information if a critical processor
failure ever occurs which allows the affected processor to be identified and removed
during the next boot. This function is provided in addition to the services to notify
customer support and service support if desired. The service processor is used to
report critical failures when AIX can not be rebooted.
The Repeat-Gard function provides the capability to remember a specific processor that
was experiencing problems during normal operation and automatically deconfigure it
during the next boot. This removes the dependency on the system to rely on the
processor to fail BIST (Built-In Self-Test) during the boot process in order to be
removed from the configuration. Processor failures may be intermittent in nature or not
re-creatable during system boot with BIST, but may fail again during normal operation
causing systems to experience a series of system failures and reboots until the faulty
processor is replaced.
You can override the default deconfiguration of the processor if so desired during the
system boot. In addition, the menu- driven interface allows the user to selectively
deconfigure or configure additional processors which are installed. This provides
customers with control over the number of processors which will be made available to
the operating system. This is particularly useful for customers performing bench-
marking runs for their applications on various configurations
88
S80 and S85 User’s Guide