The CPU errors that can result in a soft machine check are:
- System Recovery (SR): A malfunction has
occurred, but the hardware has successfully corrected or circumvented
it.
- Degradation (DG): A continuous degradation
of system performance has been detected.
The operating system does not inform the operator about the occurrence
of soft machine checks until the threshold for a given type is reached.
The default threshold set for an SR machine check is 50, and for a
DG machine check it is 1. When a threshold for a type of machine check
is reached, the system issues message IGF931E.
The MODE command allows the operator to change the threshold value
for either SR or DG machine checks, and to specify what processing
should be done when the threshold is reached.
- The operator can specify that at the threshold the CPU be disabled
for that type of machine check, that is, be put in quiet mode.
- If the MODE command specifies RECORD=ALL for a particular type
of machine check, the system does not enter quiet mode; it records
all instances of the specified type of machine check in the logrec
data set. The operating system issues message IGF931E when the number
of machine checks reaches a multiple of the threshold. For example,
if REPORT=3 is specified, message IGF931E appears after the third,
sixth, ninth, twelfth machine checks, and so on.
Numerous IGF931E messages appearing on the console might indicate
a performance degradation. In this case, the installation might want
to configure offline the processor that is experiencing the errors.
Hardware support personnel can repair the offline processor.