Pci Error Handling Mode - Hitachi Compute Blade 2000 User Manual

PCI error handling mode

This section describes PCI Error Handling Mode supported by Compute Blade 2000.
The following three PCI error handling modes are supported by Compute Blade 2000. See
Chapter 5: Server Blade Setup for the setting procedure.
Mode
PCIe Error Isolation (Default)
Legacy
OS AER
(*1) No OS is available for this feature as of March 2010.
When the EFI firmware versions "01-xx/02-xx" on

standard server blade (X55A1 model), it works as
Legacy mode without regard to the preset value of PCI
Error Handling Mode.
Other functions are supported with EFI firmware

versions "03-xx/04-xx" or later.
Description
EFI isolates PCI express unrecoverable errors, such as
disconnection of a connection cable for the I/O slot expansion
unit. Since features of the isolated PCI express due to failure
cannot recover, the system can maintain features and keep
operating by switching to a redundant path with NIC teaming
or MPIO of FC-HBA. With some serious failures, however,
notifying NMI prompts OS to reboot the same way as Legacy.
When PCI express unrecoverable errors occur, the OS
operation stops for about 500 minutes because EFI collects
failure information. Take this OS down time at the failure into
account when you design a system.
PCI express error is not isolated by EFI. When unrecoverable
PCI express failure occurs, notifying NMI prompts OS to
reboot.
OS executes Advanced Error Reporting (AER). You can set
this mode only when using OS (*1) that supports AER.
1784