Product Fault Management,
Continued
If the FOOTPRINT/DRAM experiences another error
(CRD CNT > 1), VMS sets HARD SINGLE ADDRESS
or MULTIPLE ADDRESSES along with SCRUBBED in
STATUS. Scrubbing is no longer performed; instead, pages
are marked bad. In this case, VMS logs the CRD buffer
immediately. The CRD Buffer is also logged immediately
if PAGE MAPOUT THRESHOLD EXCEEDED is set in
SYSTAT as a result of pages being marked bad. The
threshold is reached if more than one page per Mbyte of
system memory is marked bad.
NOTE
CURRENT ENTRY is zero in the Memory SBE
Reduction subpacket header if the CRD buffer
was logged, not as a result of a HARD SINGLE
ADDRESS or MULTIPLE ADDRESSES error in
STATUS, but as a result of a memory uncorrectable
ECC error shown as RELATED ERROR, or as
a result of CRD BUFFER FULL or SYSTEM
SHUTDOWN, all of which are shown under
LOGGING REASON.
VMS Event
Record
Translation
The kernel error log entries are translated from binary to ASCII
using the ANALYZE/ERROR command. To invoke the error log
utility, enter the DCL command ANALYZE/ERROR_LOG.
Syntax:
ANALYZE_ERROR_LOG [/qualifier(s)] [file-spec] [,...]
Example:
$ ANALYZE/ERROR_LOG/INCLUDE=(CPU,MEMORY)/SINCE=TODAY
The error log utility translates the entry into the traditional three-
column format. The first column shows the register mnemonics,
the second column depicts the data in hex, and the last column
shows the actual English translations.
Continued on next page
5–68
