klogd
A. Racine
aracine at u.washington.edu
Wed Feb 23 17:46:21 PST 2000
Check /etc/syslog.conf. Look for a line that has "kern." in it, like:
kern.* /var/log/kern.log
That specifies that kernel messages should to into /var/log/kern.log.
They may also be in syslog or messages. dmesg may also help. If all else
fails, you could try:
(cd /var/log/ ; grep kernel: * )
and see what shows up.
You could try using strace ('strace -p<PID>') to see what system
calls the klogd process is using. With not much happening on my machine,
the strace output from my klogd looks like:
# strace -p106
msgrcv(0, {0, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...}, 1024, -255,
MSG_NOERROR) = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) ---
alarm(60) = 0
sigreturn() = ? (mask now [])
delete_module(NULL) = 0
msgrcv(
[wait 60 seconds for the alarm]
0, {0, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...}, 1024, -255,
MSG_NOERROR) = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) ---
alarm(60) = 0
sigreturn() = ? (mask now [])
delete_module(NULL) = 0
msgrcv(
[wait again, ad infinitum]
'man syslog.conf' for more info on where syslog messages go.
You might also try restarting klogd with the -d flag. The man page says:
"This will generate LOTS of output to stderr."
Aaron
On Wed, 23 Feb 2000, R. David Whitlock wrote:
# I am unable to figure out a certain bug that someone is having on their
# machine: Once the machine is started, klogd is always sucking up all the
# CPU cycles. So root has to kill this process to get on with life. I
# realize that this is _bad_, but I'm not sure where to start debugging the
# process, such as, WHERE does the kernel error logging go, what kinds of
# things should I look for, etc? I had thought it might have been due to an
# IRQ problem, but we seem to have resolved that and klogd merrily sucks it
# all up anyway.
#
#
# Waddaya think, guys?
#
# David
#
#
#
#
More information about the Linux
mailing list