Just annotations of little "how to's", so I know I can find how to do something I've already done when I need to do it again, in case I don't remember anymore, which is not unlikely. Hopefully they can be useful to others, but I can't guarantee that it will work, or that it won't even make things worse.
Script to watch potentially misbehaving processes and kill them if needed
Posted 09-30-2016 at 10:09 AM by the dsc
Updated 10-02-2016 at 02:04 PM by the dsc (small improvements / bug fixes)
Updated 10-02-2016 at 02:04 PM by the dsc (small improvements / bug fixes)
A second adaptation of an adaptation of this script:
http://searchenterpriselinux.techtar...sses-with-Bash
The original adaptation was supposed to put any misbehaving process under cpulimit leashes, I'm not sure it's working properly. I've made a more radical implementation intended for programs that are more likely to misbehave and probably hang with high cpu, when killing them after a while seems to be a good idea.
I think it could perhaps be somewhat easily adapted to have some sort of config file to list different processes and limits, becoming somewhat similar to the "auto nice daemon" (and), but perhaps also more easily adaptable to specific requirements.
But right now you use it as "prockiller.sh process-name".
I think it won't work well with processes that spawn dozens of sub-processes, like google chrome/chromium. As always, this isn't anything intended for serious business or anything. I came up with it specifically for mpv, which was hanging with high-cpu load sometimes, to kill it if I was not around or didn't notice the fan speeding up like crazy.
http://searchenterpriselinux.techtar...sses-with-Bash
The original adaptation was supposed to put any misbehaving process under cpulimit leashes, I'm not sure it's working properly. I've made a more radical implementation intended for programs that are more likely to misbehave and probably hang with high cpu, when killing them after a while seems to be a good idea.
I think it could perhaps be somewhat easily adapted to have some sort of config file to list different processes and limits, becoming somewhat similar to the "auto nice daemon" (and), but perhaps also more easily adaptable to specific requirements.
But right now you use it as "prockiller.sh process-name".
I think it won't work well with processes that spawn dozens of sub-processes, like google chrome/chromium. As always, this isn't anything intended for serious business or anything. I came up with it specifically for mpv, which was hanging with high-cpu load sometimes, to kill it if I was not around or didn't notice the fan speeding up like crazy.
Code:
#!/bin/bash
LIMIT=${2:-2400}
while true
do
if pidof $1 ; then
info=(`ps -C "$1" -o pcpu,pid | tail -1 `)
USAGE=${info[0]%%.*}
PID=${info[1]}
USAGE=$(echo $USAGE*100 | bc -l )
echo $info
if (( USAGE > LIMIT-800 ))
then
USAGE1=$USAGE
PNAME1=$PNAME
echo $1 -- usage $USAGE \(pid $PID\)
sleep=2
sleep $sleep
info2=(`ps -C "$1" -o pcpu,pid | tail -1 `)
USAGE2=${info2[0]%%.*}
USAGE2=$(echo $USAGE2*100 | bc -l )
PID2=${info2[1]}
echo $1 -- usage $USAGE \(pid $PID2\) ... limit: $LIMIT
[ ! $PID = $PID2 ] && highload=0
((USAGE2 > LIMIT )) && [ $PID = $PID2 ] && highload=$((highload+1)) && echo "high cpu load for $highload times!"
((highload==5)) && if [[ ! "$*" == *"-dry-run"* ]] ; then kill -kill $PID ; fi && highload=0 && echo "=============== had to kill $1 ============== "
((USAGE2 < LIMIT )) && highload=$((highload-1)) ; ((highload<0)) && highload=0
else
sleep=10
fi
sleep ${sleep}
else
echo no $1 process yet...
sleep 10
fi
done
Total Comments 0
