Periodic diagnostics

Edit online

Periodic testing of the disk drives and battery are enabled by default on systems that support these tests. The disk diagnostics perform disk error log analysis on all disks. The battery test checks the real-time clock and NV-RAM battery.

Periodic diagnostics are performed in different ways, depending on the diagnostic version. Use the Periodic Diagnostics task to change the test times or to add other resources to the list.

Processors that are dynamically removed from the system will also be removed from the periodic test list. Processors that are dynamically added are automatically added to the periodic test list.

Periodic Diagnostics supports only processors. Periodic Diagnostics automatically analyzes the error log and reports the errors. Periodic Diagnostics supports processor diagnostics on systems that own the processor. However, some hardware platforms own the processors and provide testing instead of AIX®, so Periodic Diagnostics on those processors is not supported. Errors on the processors owned by the platform are logged by the platform and analyzed by automatic error log analysis.

Support is determined at run time from the characteristics of the hardware configuration. Periodic Diagnostics determines it is not supported when processors do not have physical location codes, which are created by platform firmware and passed to the operating system.

If processors have a physical location code, then the partition owns the processor and Periodic Diagnostics is supported. If there is no physical location code for a processor, it is a shared or virtual processor and Periodic Diagnostics is not supported.

Technical Description

The Diagnostic daemon diagd executes once the bos.diag diagnostic package is installed. The diagd looks for customized entries in CDiagAtt odm database to determine which devices to run at which times. The database is built when diagnostics are run or the Periodic Diagnostic Service Aid is run to change run times for devices. If the database has no entries (for example, when diagnostics have never been run), then default times are given to the ioplanar battery test and disk drives. The following is an example of CDiagAtt entries.


CDiagAtt->attribute = p_test_time
CDiagAtt->value = 9999 Do not test
        = 0400 Test at 4AM

The diagd sets a timer to wake up at the next scheduled time to run. Once diagd wakes up, the script /usr/lpp/diagnostics/bin/diagela is executed with the -t flag.

diagela checks the PDiagAtt->test_mode bit for the device to determine whether that device should be tested in this mode. If the bit is not set, diagela does not test the device. If the bit is set, diagnostics are run on the device with the -e (ELA) flag set.

Platform firmware, rather than Periodic Diagnostics, is used to test memory and processors on some systems. AIX® still reports errors on theses resources, but firmware does the actual testing. If the diagd determines that there are no resources to test periodically, it will terminate.

Limitation

The IBM® System p5® does not support AIX® processor diagnostics and Periodic Diagnostics in any version of AIX®.