IBM RS/6000 SP, Problem Determination Manual

Page: 182 / 316

This soft copy for use by IBM employees only.
/var/adm/ras/errlog Stores instances of errors and failures
encountered by the system.
/var/adm/ras/errtmplt Contains the Error Record Template
Repository.
The /dev/error and /dev/errorctl special files support the logging of error events.
Minor device 0 (zero) of the error special file is the interface between processes
that log error events and the errdemon daemon. Error records are written to
/dev/error by the errlog library routine and the errsave kernel service. The error
special file timestamps each error record entry.
errdemon opens /dev/error for reading. Each read retrieves an entire error
record. The format of the error records are described in the header file
/usr/include/sys/err_rec.h.
Each time a error is logged, the error ID, the resource name, and the
timestamps are recorded in Non Volatile Random Access Memory (NVRAM) so
that in the event of a system crash, the last logged error is not lost. When the
error file is started, the last error entry is retrieved from NVRAM.
The standard device driver interfaces (open, close, read, and write) are provided
for the error file. The error file has no
ioctl functions. The ioctl function
interface for the error special file is provided by the errorctl special file. This
interface supports stopping the error logging system, synchronizing the error
logging system, and querying the status of the error special file.
6.2 SP Error Logging
The RS/6000 SP uses both the AIX Error Logging facilities and the BSD syslog,
as well as a number of function-specific log files to record error events on each
node.
Commands and SMIT panels are available to perform general log management.
For managing Syslog and AIX Error Log, it is necessary to install the ssp.sysman
option.
Error logging is the writing of information to persistent storage to be used for
debugging purposes. This type of logging is for subsystems that perform a
service or function on behalf of an end user. The subsystem does not
communicate directly with the end user and therefore needs to log events to a
storage location. The events that are logged are primarily error events.
Error logging for the SP uses BSD syslog and AIX Error Log facilities to report
events on a node basis. The System Monitor, the High Performance Switch, SP
Switch, and the Resource Manager components use this form of error logging.
Error log entries include a DETECTING MODULE string that identifies the
software component, module name, module level, and line of code or function
that detected the event that was logged. The information is formatted depending
on the logging facility the user is viewing. For example the AIX Error Log facility
information appears as follows:
DETECTING MODULE
LPP=<LPP name> Fn=<filename> <ID_level_of_the_file> L#=<line number>
162
SP PD Guide