monitoring/remove_zombies_user.sh

#!/usr/bin/env bash

working_dir="$(dirname $(realpath $0))"

# -------------
# - Some Variables
# -------------

LOCK_DIR="/tmp/$(basename $0).LOCK"

KILL_SIGNAL=HUP

#CHECK_USER="$*"


# -------------
# - Some functions
# -------------

usage() {

   
   [[ -n  "$1" ]] && error "$1"

   
   [[ $terminal ]] && echo -e "
\033[1mUsage:\033[m

   $(basename $0) -u <user>

\033[1mDescription\033[m

   Searches for zombie (defunct) child prozesses owned by a given user. If such
   processes exists, the script tries to kill them by sending a HUP signal to
   the parent process.

   A user must be given using option '-u'.

\033[1mOptions\033[m

   -h
      Prints this help.

   -u <user>
      Searches for zombie (defunct) prozesses owned by this user.

\033[1mExample:\033[m
	
   Search for (and delete) defunct child prozesses owned by user 'sympa''

      $(basename $0) -u sympa

"

   clean_up 1

}

clean_up() {

   # Perform program exit housekeeping
   rm -rf "$LOCK_DIR"
   exit $1
}

trim() {
    local var="$*"
    var="${var#"${var%%[![:space:]]*}"}"   # remove leading whitespace characters
    var="${var%"${var##*[![:space:]]}"}"   # remove trailing whitespace characters
    echo -n "$var"
}

fatal(){
   echo ""
   if $terminal ; then
      echo -e "   [ \033[31m\033[1mFatal\033[m ]: $*"
      echo ""
      echo -e "   \033[31m\033[1mScript was interupted\033[m!"
   else
      echo "[ Fatal ]: $*"
      echo ""
      echo "              Script was terminated...."
   fi
   echo ""
   clean_up 1
}

error (){
   echo ""
   if $terminal ; then
      echo -e "   [ \033[31m\033[1mError\033[m ]: $*"
   else
      echo "[ Error ]: $*"
   fi
   echo ""
}

warn (){
   echo ""
   if $terminal ; then
      echo -e "   [ \033[33m\033[1mWarning\033[m ]: $*"
   else
      echo "[ Warning ]: $*"
   fi
   echo ""
}

info (){
   echo ""
   if $terminal ; then
      echo -e "   [ \033[32m\033[1mInfo\033[m ]: $*"
   else
      echo "[ Info ]: $*"
   fi
   echo ""
}
info_terminal (){
   if $terminal ; then
      echo ""
      echo -e "   [ \033[32m\033[1mInfo\033[m ]: $*"
      echo ""
   fi
}

ok (){
   echo ""
   if $terminal ; then
      echo -e "   [ \033[32m\033[1mOk\033[m ]: $*"
   else
      echo "[ Ok ]: $*"
   fi
   echo ""
}


# -------------
# - Check some prerequisites
# -------------

# - Is this script running on terminal ?
# -
if [[ -t 1 ]] ; then
   terminal=true
else
   terminal=false
fi

if [[ -z "$(which basename)" ]]; then
   fatal 'It seems "basename" is not installed, but needed!'
fi

if [[ -z "$(which realpath)" ]]; then
   fatal 'It seems "realpath" is not installed, but needed!'
fi



# -------------
# - Jobhandling 
# -------------

# - If job already runs, stop execution..
# -
if mkdir "$LOCK_DIR" 2> /dev/null ; then

   ## - Remove lockdir when the script finishes, or when it receives a signal
   trap clean_up SIGHUP SIGINT SIGTERM

else

   datum="$(date +"%d.%m.%Y %H:%M")"

	fatal "A previos instance of that script \"$(basename $0)\" seems already be running."

   exit 1

fi


# -------------
# - Read Commanline Arguments
# -------------

while getopts hu: opt ; do
	case $opt in
	 	h) usage ;;
	 	u) CHECK_USER=$OPTARG ;;
   	*) usage
	esac
done

if [[ -z "$CHECK_USER" ]]; then
	usage "No user (option -u) given"
fi

shift $(expr $OPTIND - 1)
[[ "$#" -gt 0 ]] && usage "Wrong number of arguments given!"


# -------------
#  Checking ..
# -------------

if $terminal ; then
   echo ""
   echo -e "   \033[1mCheck for zombie (defunct) child prozesses owned by a given user\033[m"
   echo "   ================================================================"
fi


if $(ps -Ao"stat,user,command" | grep "${CHECK_USER}" | grep -v grep | grep -e '^[zZ]' > /dev/null 2>&1)
then

   warn "Found Zombie Processes '${CHECK_USER}' on '$(hostname -f)':"

   # - Print out the concerning lines frp 'ps' output.
   # -
   ps -Ao"stat,user,pid,ppid,command" | grep "$(basename ${CHECK_USER})" | grep -v grep | grep -e '^[zZ]'

   membefore=$(cat /proc/meminfo | grep Committed_AS | awk '{print $2}')

   info "Trying a graceful kill to the concerning parents with signal '$KILL_SIGNAL' .."

	# - Process ID(s) of the zombie process(es)
	# -
	ZOMBIE_PIDS="$(ps -Ao"stat,user,pid,ppid,command" | grep "${CHECK_USER}" | grep -v grep | grep -e '^[zZ]' \
      | awk '{ print $3 }' | sort -u)"

	# - Sending the kill signal ($KILL_SIGNAL) to the parent process ID(s)
	# -
   ps -Ao"stat,user,pid,ppid,command" | grep "${CHECK_USER}" | grep -v grep | grep -e '^[zZ]' \
		| awk '{ print $4 }' | sort -u | xargs --no-run-if-empty kill -$KILL_SIGNAL

   if [[ "$?" -eq 0 ]];then

		declare -i count=0
		for _PID in $ZOMBIE_PIDS ; do
			if [[ $count -eq 1 ]]; then
				KILLED_PIDS="$_PID"
			else
				KILLED_PIDS="$KILLED_PIDS $_PID"
			fi
			((count++))
		done

      sleep 10

      ok "Cleaning up zombie processes owned by user '${CHECK_USER}' seems succsessfully finisched"

		ok "$count zombie process(es) with PID(s) '${KILLED_PIDS}' removed from process list"

      memafter=`cat /proc/meminfo | grep Committed_AS | awk '{print $2}'`

      info "`expr $membefore - $memafter` KB RAM has been freed"
   else
      error "Cleaning up zombie processes owned by user '${CHECK_USER}' failed!"
   fi
else
   info_terminal "No zombie prossses owned by user '${CHECK_USER}' found."
fi

if $terminal ; then
   echo ""
fi
clean_up 0