check_systemdunit

#!/bin/bash
# ================================================================================
#
# CHECK A SINGLE SYSTEMD SERVICE
#
# -------------------------------------------------------------------------------
# 2023-09-05  v1.0  <axel.hahn@unibe.ch>
# 2020-09-06  v1.1  <axel.hahn@unibe.ch>  add params -s, -l
# 2020-09-06  v1.2  <axel.hahn@unibe.ch>  add param -r to use a regex
# 2020-09-18  v1.3  <axel.hahn@unibe.ch>  fix: list units --all to show all stopped units
# 2020-09-18  v1.4  <axel.hahn@unibe.ch>  replace pipes in systemctl status output
# 2020-10-20  v1.5  <axel.hahn@unibe.ch>  remove special chars from systemd status
# 2020-10-23  v1.6  <axel.hahn@unibe.ch>  handle units with multiple instrances
# 2020-11-03  v1.7  <axel.hahn@unibe.ch>  autodetect multiple instances without @ char in unit name
# ================================================================================

. $( dirname $0 )/inc_pluginfunctions

export self_APPVERSION=1.7

# ----------------------------------------------------------------------
# FUNCTIONS
# ----------------------------------------------------------------------

# show help text
function showHelp(){
    local _self; _self=$(basename $0)
cat <<EOF
$( ph.showImlHelpHeader )

Check a unit using systemctl status.

The status is "unknown" if the command systemctl is not found.
The status is "critical" if the service does not exist or is not running.

When checking a service with multiple instances you get status "warning"
if not all instances are active.

SYNTAX:
  $_self [-h|-l|-s|-r] UNIT

OPTIONS:

  -h     this help
  -l     list all units
  -s     list service units
  -r     handle UNIT as a regex and search for a single unit. A uniq regex
         is needed to match a single unit. The initial idea was to match a
         servie that has different service names on differen os eg.
         - apache2 vs httpd
         - mysld vs mariadb

  UNIT   Name of a unit - see output of 'systemctl' 

EXAMPLES:

  $_self -s
         list all existing services. For a unit check you need to add the name
         in the 1st column.

  $_self nginx.service
         show status of nginx webservice

  $_self -r "(apache2|httpd)\.service"
         Detect name of Apache httpd service by given regex and show status
         of the found service name

  $_self something@*
         Check if all instances of a service "something@.service" are active.
         If not all instances are running you get status "warning", if none 
         is running "critical".

  $_self something@2.service
         Check instance 2 of service "something".

EOF
}

# get the list of units
function _getUnits(){
  systemctl --no-legend --no-pager --all
}

# ----------------------------------------------------------------------
# MAIN
# ----------------------------------------------------------------------

ph.hasParamoption "h" "$@"; bOptHelp=$?

if [ $bOptHelp -eq 0 -o $# -eq 0 ]; then
    showHelp
    exit 0
fi

ph.require "systemctl"

# --- list all units
if ph.hasParamoption "l" "$@" ; then
  echo "List of all systemd units:"
  echo
  _list=$( _getUnits )
  for mytype in $( awk '{ print $1 }' <<< "$_list" | rev| cut -f 1 -d '.' | rev | grep -v '[^a-z]' | sort -u )
  do
    echo "---------- $mytype"
    grep "\.${mytype}" <<< "$_list"
    echo
  done
  exit 0
fi

# --- list service units
if ph.hasParamoption "s" "$@" ; then
  echo "List of service units:"
  echo
  systemctl --no-legend --no-pager --type service
  exit 0
fi

# --- find a service by regex
if ph.hasParamoption "r" "$@" ; then
  _regex="${2}"
  _list=$( _getUnits | awk '{ print $1 }' | sort )
  _unit=$( grep -E "$_regex" <<< "$_list" )

  if [ -z "$_unit" ]; then
    ph.setStatus critical
    ph.status "REGEX '$_regex' does not match any existing unit"
    ph.exit
  fi
  
  if [ $( wc -l <<< "$_unit" ) != "1" ]; then
    ph.setStatus critical
    ph.status "REGEX '$_regex' matches on multiple units on this system:"
    echo "$_unit" | nl
    echo "You need to use a uniq regex."
    ph.exit
  fi
else
  _unit="${1}"
fi

# --- check given unit

_data=$( systemctl --no-pager -l status "${_unit}" 2>&1 | tr '|' ':' )

_status=$( echo "$_data" | head -1 | sed "s#^[^a-zA-Z]*##g" )
_out=$( echo "$_data" | sed -n "2,\$p")

# get "running" instances
typeset -i _iActive;    _iActive=$( grep -c "Active: active (running) " <<< "${_data}" )

if [ $_iActive -eq 0 ]; then
  ph.setStatus critical
else
  typeset -i _iInstances; _iInstances=$( grep -c "Loaded: " <<< "${_data}" )
  if [ $_iInstances -gt 1 ]; then

    _status="$_iActive of $_iInstances ${_unit} units are active"
    _out="$_data"

    if [ $_iInstances -ne $_iActive ]; then
      ph.setStatus warning
    fi
  fi

fi
ph.status "${_status}"

echo "$_out" | tr -d '└├●' | tr '─' '-' 

ph.exit

# ----------------------------------------------------------------------