Fault Management

When a server component fails, error telemetry is either captured via the BIOS or is monitored by the Oracle ILOM SP. Oracle ILOM consumes error telemetry from both sources and provides diagnosis in the form of a fault event. The fault event is stored in the Oracle ILOM event log as a fault message. You can use either the Oracle ILOM web interface or the command-line interface (CLI) to manually clear faults.

This section includes the following topics. The first four topics describe how to examine and clear faults, while the last topic provides reference information for sensors and indicators.

• Determining Faults

• Clearing Faults

• Components With No Fault Diagnosis

• Viewing Sensors Using IPMItool

Determining Faults

When a system fault occurs, you can view system indicators and use the Oracle ILOM CLI or web interface to determine the fault:

• LEDs – The Service Required LED will always be illuminated, and the component or subsystem-specific Service LED will be illuminated when applicable.

• Oracle ILOM CLI – Examine fault messages in the Oracle ILOM event log or see a fault summary.

  For example:

  • To view the Oracle ILOM event log, log in to the Oracle ILOM CLI and type: show /SP/logs/event/list

  • To view a fault summary, log in to the Oracle ILOM CLI and type: show /SP/faultmgmt

• Oracle ILOM web interface – Examine fault messages in the Oracle ILOM event log or see a fault summary.

  For example:

  • To view the Oracle ILOM event log, log in to the Oracle ILOM web interface and select: System Monitoring --> Event Logs

  • To view a fault summary, log in to the Oracle ILOM web interface and select: System Information --> Fault Management

Clearing Faults

The procedure for clearing a fault differs depending on the type of component.

1. Customer-replaceable units (CRUs) that are hot-swappable and are monitored by the SP will have their faults cleared automatically when the failed component is replaced and the updated status is reported as deasserted.

2. CRUs and field-replaceable units (FRUs) that have a FRUID container with identity information will have their faults cleared automatically when the failed component is replaced, as the SP is able to determine when a component is no longer present in the system.

3. CRUs and FRUs that are not hot-swappable or lack a FRUID container with identity information will not have their faults cleared automatically.

You can use the Oracle ILOM web interface or the command-line interface (CLI) to manually clear faults. For information on how to use the Oracle ILOM web interface or the CLI to clear server faults, see the Oracle ILOM 3.0 Documentation Library at:

http://www.oracle.com/pls/topic/lookup?ctx=ilom30

The following types of faults are diagnosed by the Oracle ILOM SP:

• Environmental events – Fan modules, power supplies, ambient temperature, AC power loss, and chassis intrusion switch

• Memory Reference Code (MRC) errors and warnings – Memory initialization and population

• I/O Hub (IOH) uncorrectable error events – Motherboard

• Memory ECC uncorrectable and correctable events – Memory DIMMs

• CPU uncorrectable error events – Processor

• Boot progress events – Power-on, power-off, IPMI, MRC, QPI, BIOS, setup, and boot retries

• Service Processor error events – ILOM

Component Fault Events  lists the server component faults that are persistent after a system cold boot and the action to clear the fault.

Table 3-1 Component Fault Events 

In addition to the above faults, the following fault does not require replacement of a faulty part; however, user action is needed to clear it:

fault.security.integrity-compromised@/sys/sp

This fault is generated when the server's top cover is removed while the AC power cords are still connected to the power supply, that is, power is not completely removed from the server. To clear this fault, replace the server's top cover and either reboot the server's SP or remove the AC power cords, and then reconnect the power cords.

Components With No Fault Diagnosis

Certain Sun Server X2-4 components do not provide a mechanism to diagnose faults. These include:

• Disk backplane

• DVD player

• Disk drive

• Power supply backplane

• Lithium battery for host and SP real-time clocks

Viewing Sensors Using IPMItool

Sun Server X2-4 sensors can be viewed using IPMItool. For information and instructions for viewing sensors using IPMItool, see the Oracle Integrated Lights Out Manager (ILOM) 3.0 Protocol Management -- SNMP, IPMI, CIM, WS-MAN Guide.