 |
» |
|
|
|
NAMEkl — control kernel logging SYNOPSIS/usr/sbin/kl
-e
[-q
qsize]
[-s
fsize]
[-w
(on|off)]
[-l
{d|e|w|i}
{subsys_name...|
all}] /usr/sbin/kl -d /usr/sbin/kl -l
{d|e|w
|i}
{subsys_name...|
all} /usr/sbin/kl -w
{on
[-s
fsize ] |
off} /usr/sbin/kl -s fsize /usr/sbin/kl -i /usr/sbin/kl -p filename
[-w on
[-s
fsize ]] /usr/sbin/kl -q qsize DESCRIPTIONThe
kl
command controls the operation of the Kernel Logging facility.
Kernel Logging is a high-availability feature
that gives system administrators
the ability to collect the information necessary to diagnose
problems with the HP-UX kernel
while the system is running.
kl
is used to specify the levels of events to be logged and the
kernel subsystems that will write messages to memory or disk.
kl
also provides
for managing the contents of the logfile in memory and on disk. At startup, Kernel Logging determines its default configuration
by reading the file
/etc/nettlgen.conf.
See
nettlgen.conf(4)
for an explanation of the file format.
The
kl
command permits only temporary changes
to the default Kernel Logging configuration
without having to stop and restart Kernel Logging facility.
Note that any values you specify on the
kl
command line do not modify the contents of the
/etc/nettlgen.conf
file.
To make permanent changes to the values in the
/etc/nettlgen.conf
file, run
nettlconf(1M). Only users with appropriate privileges (root) can invoke the
kl
command to control
the Kernel Logging facility. Optionskl
recognizes the following options, which can be used only in the combinations
indicated in the
SYNOPSIS
section.
All options and keywords are case-sensitive.
- -e
Enable the Kernel Logging facility and start up default logging as
defined in the file
/etc/nettlgen.conf. If the
-l
option is used in conjunction with the
-e
option, the
-l
option must be specified as the last option on the command line. - -d
Disable the Kernel Logging facility.
Once this option is issued,
Kernel Logging stops accepting logging calls from the kernel subsystems. - -w {on|off}
Turn on/off write-to-disk logging.
If write-to-disk logging is
enabled, log messages residing on the log queue in memory
are written to disk and removed from the queue. The name of the log file on disk is formed by adding the suffix
.KLOG0
to the log file name specified in the
/etc/nettlgen.conf
file.
If the log file (including suffix) already exists,
one of the following events takes place:
If the existing file contains messages logged during the system run when a panic
occurred, the file will be preserved (in a new location).
See the USAGE section for more information on how this situation is handled. Otherwise, the existing file and whatever
information it contains will be lost;
that is, the contents of the file are overwritten with new kernel logging data.
When write-to-disk logging is turned off, messages are not
written out to disk, but continue to be collected in memory (circular buffer). - -l { d | e | w | i } { subsys_name ... | all }
Modify the level of log messages to be captured for the specified
subsystem(s). The
subsys_name
argument is a subsystem name specified in the file
/etc/nettlgen.conf.
The list of available subsystem names can be
obtained using the command
nettlconf -KL -status.
The keyword
all
changes the logging level for all subsystems specified in the file
/etc/nettlgen.conf. The following table describes the classes of messages
that can be logged
The following table identifies the classes of messages that are
captured at each log level
Note that, although the log level is specified as a single keyword, messages are
logged according to the following rule: if level
x
is specified,
then all messages whose severity is greater than or equal to the
severity of class
x
will be logged. The order of severity is as follows: Disaster (the most severe),
Error, Warning, Informative (the least severe). - -s fsize
Set the size of the two log files used to store logged messages
when write-to-disk is enabled.
The maximum
fsize
is 1 gigabyte;
the minimum
fsize
is 16 kilobytes. When write-to-disk is started,
the default log file size is taken from the file
/etc/nettlgen.conf.
The
-s
option allows you to modify the size of the log file
without stopping write-to-disk operations. Note that, when setting the file size,
you can specify a numeric value followed by the character
'M'
or
'K',
which indicates that
fsize
is being defined
in units of megabytes or kilobytes, respectively.
This suffix character is case-insensitive. - -p filename
Take a picture (snapshot) of the current contents in memory of the kernel
logging buffer.
This option causes all messages residing in the
log queue to be dumped to
filename
and removed from the queue. Note that taking a picture (snapshot) and write-to-disk are mutually
exclusive operations, because there is no reason to take a picture of
Kernel Logging messages if write-to-disk component already writes them to disk. If the
-w on
option is specified in conjunction with the
-p
option,
write-to-disk logging will be started
immediately after the picture is taken. - -q qsize
Set the size of the Kernel Logging queue.
qsize
indicates the number of buffers of size 8 kbytes each.
The maximum
qsize
is 1024 buffers; the minimum
qsize
is 8 buffers.
At startup, this value is taken from the file
/etc/nettlgen.conf. - -i
Report information about the status of the Kernel Logging facility. Information returned by the
-i
option includes:
kernel logging facility is on or off. write-to-disk logging is on or off. picture (snapshot) tool is on or off. current size of the Kernel Logging queue. number of messages currently held in the queue. name of the log file used by write-to-disk. maximum file size of the log file used by write-to-disk,
together with the minimum file size available for the current session of
write-to-disk. number of messages not written to file (could be due to the lack of memory or
small size of the circular buffer). list of all subsystems currently specified in the
/etc/nettlgen.conf
file and the associated message class(es) logged for each subsystem.
USAGEWhen to Use Kernel LoggingMission critical systems should have KL always enabled.
Failing to enable Kernel Logging causes diagnostic information
about any suspicious events that might occur on the system to be lost.
The recommended classes to capture are Disaster, Error and Warning. Use
kl -l w all
command to do so.
To minimize Kernel Logging's impact on a running system,
use the
kl -l e all
command to set all kernel subsystems
to capture error-level log messages only.
Log File ManagementThe write-to-disk facility uses two files to hold logging information.
The base name of the log files is specified in
/etc/nettlgen.conf.
The default base log file name is
/var/adm/kl;
see
nettlgen.conf(4).
The most current data is always in the file with suffix
.KLOG0.
If the size of the
.KLOG0
file reaches a user-defined maximum, Kernel Logging renames the
.KLOG0
file to
.KLOG1,
overwriting the previous contents of the
.KLOG1
file, then continues writing messages to the
.KLOG0
file.
(Specify maximum log file size in
/etc/nettlgen.conf
or use the
-s
option.) The Kernel Logging facility has a protection feature for saving old log files.
When write-to-disk starts and encounters old log files that contain messages
collected during a prior run of the system when a panic occurred (thus, the log
files may contain important information about the panic), then log files
are not overwritten.
Instead, write-to-disk first attempts to move the old log files
to the default crash directory
(typically
/var/adm/crash/crash.ID,
where
ID
is a numeric counter).
If the move fails, then write-to-disk
tries to rename the log files with the prefix
OLD,
allowing the files to remain in the same directory.
For example, if write-to-disk finds an old log file named
/var/adm/kl.KLOG0,
it first tries to move the file into the default crash directory.
If this move is not successful, write-to-disk then tries to move the
file to
/var/adm/OLDkl.KLOG0. If both attempts to save old log files fail,
the write-to-disk component is not started.
To preserve the messages in the old log files,
do one of the following:
If possible, eliminate the conditions
that caused both attempts to move the old log files to fail. Manually move old log files. Using the
nettlconf
command, modify the file
/etc/nettlgen.conf
to specify a different log file name for write-to-disk logging.
Once the problem is corrected, start write-to-disk again.
If successful, the Kernel Logging facility will notify you about the move of the
old log files. RETURN VALUEkl
exits with one of the following values:
- 0
Operation was successful - 1
kl
command aborted due to error(s)
EXAMPLES- 1.
Enable the default Kernel Logging facility. kl -e - 2.
Display the information about the Kernel Logging facility. kl -i - 3.
Change level to log disaster, error and warning messages for all subsystems. kl -l w all - 4.
Request a picture (snapshot) of the collected log messages. kl -p kl_snap.kl - 5.
Turn on default write-to-disk logging. kl -w on - 6.
Change the maximum size of the current log file to 512KB. kl -s 512K - 7.
Turn off write-to-disk logging. kl -w off - 8.
Disable the Kernel Logging facility. kl -d - 9.
Enable on the Kernel Logging facility with the following parameters:
queue size is 1000, write-to-disk component is on,
maximum file size is 4 megabytes,
and log level for the KL_PM subsystem is Warning. kl -e -q 1000 -w on -s 4M -l w KL_PM Note that
-l w KL_PM
changes the initial level of the KL_PM subsystem only.
This means that the initial levels of the rest of the subsystems
specified in the
/etc/nettlgen.conf
file remain unchanged. - 10.
Take a picture (snapshot) and start write-to-disk logging with a
maximum file size of 128K. kl -p -w on 128K
AUTHORkl
was developed by HP in partnership with NEC. FILES- /dev/kernlog
Kernel log pseudo-device file. - /etc/nettlgen.conf
NetTL and KL subsystem configuration file. - /var/adm/kl.KLOG0
Default log files as specified in /etc/nettlgen.conf - /var/adm/kl.KLOG1
|
