As the fam maintainer (or is it the debian package maintainer? I don't now) did
not fix that bug
for several years now I gave up waiting for a solution.
I am not a linux guru at all and this solution may be kinda home-style cooking,
but for me it works.
1. Create a file named famdguard in a directory of your choice <YourDir> with
the following content:
--- snip --------------------------------------------------------
#!/bin/sh
set -x
COUNT=0
# init the loop control variable
USAGE=100
# Check three times if the famd process is above xx%
# Restart fam if that is the case.
# Set MAXUSAGE to 80 for a single core machine
# Set it to 45 for a dual core machine (one processor completely used)
# Set it to 22 for a quad core machine (one processor completely used)
MAXUSAGE=45
while [ 0$USAGE -gt $MAXUSAGE ]; do
# get the cpu usage for the fam process
USAGE=`ps -eo pcpu,pid -o comm= | sort -k1 -n -r | grep famd | awk '{ print
$1 }'`
# remove the fractional part of the cpu usage value
USAGE=${USAGE%%.*}
# Increment COUNT if fam's cpu usage is above MAXUSAGE
[ 0$USAGE -gt $MAXUSAGE ] && COUNT=`expr $COUNT + 1`
if [ $COUNT -gt 2 ]; then
# Three times high CPU usage. Stop and Restart famd
/etc/init.d/fam stop
# Check, if it was successfull or try to kill famd
ps -e | grep famd
[ $? -eq 0 ] && killall -9 famd
# Restart fam
/etc/init.d/fam start
exit
fi
# Wait some time before the next loop
sleep 30
done
--- snap --------------------------------------------------------
This file checks for the CPU usage of fam.
If it is above the MAXUSAGE it checks two times more with a delay of 30
seconds.
If CPU usage is above MAXUSAGE 3 times the fam deamon is stopped and started
again.
2. make that file executable
$ chmod +x <YourDir>/famdguard
3. add the following line into your /etc/crontab to perform the check every 10
minutes
# Check for famd running wild
*/10 * * * * root /<YourDir>/famdguard
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
