A sample script to monitor systemd service units forever.

systemd-svc-mon.sh

#! /bin/bash
#
# A sample script to monitor systemd service units forever.
#
# Requirements: timeout and sleep in coreutils, uuidgen in utl-linux and systemd
#
# Author: Satoru SATOH <[email protected]>
# License: MIT
#
set -e

USAGE="Usage: $0 SVC_NAME_0 [SVC_NAME_1 [SVC_NAME_2 [...]]]"

INTERVAL=10m   # How long to sleep. See also sleep(1).
TIMEOUT=3m     # How long to wait for 'systemctl is-active ...' finished. See also timeout(1).

HELP_TXT="$USAGE
Options:
  -I INTERVAL  Interval to monitor services [${INTERVAL}].
               The format is same as sleep(1).
  -T TIMEOUT   Timeout to wait for getting the status of each service [${TIMEOUT}].
               The format is same as timeout(1).

  -v           Verbose mode
  -h           Show this help"

function show_help () {
  cat << EOH
${HELP_TXT:?}
EOH
}

function log () {
  logger --journald << EOM
PRIORITY=err
MESSAGE_ID=$(uuidgen)
MESSAGE=$@
EOM
}

function vecho () { :; }  # Do nothing.

while getopts I:T:vh OPT
do
  case "${OPT}" in
      I) INTERVAL=${OPTARG}
         ;;
      T) TIMEOUT=${OPTARG}
         ;;
      v) vecho () { echo "$@"; }
         ;;
      h) show_help; exit 0
         ;;
      \?) echo '[Error] Invalid option[s] and/or arguments!' > /dev/stderr; exit 1
         ;;
  esac
done
shift $((OPTIND - 1))
test $# -gt 0 || { echo $USAGE; exit 1; }

SVC_NAMES="$@"

while true; do
  for svc in ${SVC_NAMES}; do
    vecho "[Info] Checking if the service ${svc} is running..."
    timeout ${TIMEOUT:?} systemctl is-active ${svc}.service >/dev/null 2>/dev/null || \
      { log "${svc} does not look active (running)"; exit $?; }
  done
  sleep ${INTERVAL:?}
done

# vim:sw=2:ts=2:et: