Scripts nagios - exemples

Exemples de scripts nagios.

Load average pour tous les Unix

Il existe de nombreux scripts qui surveillent le load average sur unix, mais tous ceux que j'ai trouvé lèvent une alerte sur une valeur donnée, sans lien avec le nombre de cpus. Or le load average n'a de sens que comparé au nombre de cpus. J'ai donc développé ce script qui surveille le load en fonction du nombre de cpus. Les valeurs optionnelles qu'on lui passe sont un pourcentage par rapport au nombre de cpus. Les valeurs par défaut ont été choisies pour s'adapter à peu près partout.

#!/usr/bin/env python

"""
Nagios plugin to check load average on GNU/linux or Solaris servers.

This script checks the load average in comparison to the number of cpus.
There is a critical alarm if load 1min >= 100% of cpu number (adjustable with "-c" option).
There is a warning alarm if load 1min >= 80% of cpu number (adjustable with "-w" option).
If you want to set an alarm to 70%, for example, you have to write "0.7", not "70%".

Example:
%prog -w 0.8,0.8,0.8 -c 1.2,1.2,1.2 # Warning = 80% of cpu number and critical = 120%
%prog -w 2,1.5,1 # Warning = 200% of cpu number for load 1 min, 150% for load 5 min and 100% for load 15 min.

Result example:
OK - nb cpus: 4; load average: 0.02, 0.03, 0.0; warning: 6.0, 4.8, 4.0; critical: 10.0, 8.0, 6.0|nb_cpus=4;load_1=0.02;load_5=0.03;load_15=0.0

frederic.menard-prestataire@laposte.fr
"""

import sys, optparse, os
from optparse import OptionParser

VERSION="2.0 (2015-01-13)"


### Parse commandline options
Usage = "Usage: %prog [options]\
\nExample: %prog -w 0.8,0.8,0.8 -c 1.2,1.2,1.2 # Warning = 80% of cpu number and critical = 120%"
parser = OptionParser(usage=Usage,version="%prog version " + VERSION)
parser.add_option("-w", "--warning",
    action="store", type="string", dest="warn_threshold", help="Warning threshold in % of cpu number for load 1,5,15 min", default="1,0.8,0.7")
parser.add_option("-c", "--critical",
    action="store", type="string", dest="crit_threshold", help="Critical threshold in % of cpu number for load 1,5,15 min", default="1.2,1,0.8")
(options, args) = parser.parse_args()


### Extracting options into variables
warning  = tuple(options.warn_threshold[0:len(options.warn_threshold)].split(','))
warn_1   = float(warning[0])
warn_5   = float(warning[1])
warn_15  = float(warning[2])
critical = tuple(options.crit_threshold[0:len(options.crit_threshold)].split(','))
crit_1   = float(critical[0])
crit_5   = float(critical[1])
crit_15  = float(critical[2])


### Variables
try:
    nb_cpus = int(os.sysconf('SC_NPROCESSORS_ONLN'))
except (AttributeError, ValueError):
    pass
try:
    import multiprocessing
    nb_cpus = multiprocessing.cpu_count()
except (ImportError, NotImplementedError):
    pass
warn_1   = warn_1 * nb_cpus
warn_5   = warn_5 * nb_cpus
warn_15  = warn_15 * nb_cpus
crit_1   = crit_1 * nb_cpus
crit_5   = crit_5 * nb_cpus
crit_15  = crit_15 * nb_cpus
load_avg = os.getloadavg()
load_1   = load_avg[0]
load_5   = load_avg[1]
load_15  = load_avg[2]


### Tests
if load_1 > crit_1 or load_5 > crit_5 or load_15 > crit_15:
    status = ("CRITICAL",2)
elif load_1 > warn_1 or load_5 > warn_5 or load_15 > warn_15:
    status = ("WARNING",1)
elif load_1 >= 0 and load_5 >= 0 and load_15 >= 0:
    status = ("OK",0)
else:
    status = ("UNKNOWN",3)


### Display results
nb_cpus = str(nb_cpus)
w1      = str(warn_1)
w5      = str(warn_5)
w15     = str(warn_15)
c1      = str(crit_1)
c5      = str(crit_5)
c15     = str(crit_15)
l1      = str(load_1)
l5      = str(load_5)
l15     = str(load_15)

STATUS = status[0] +" - nb cpus: "+nb_cpus+"; load average: "+l1+", "+l5+", "+l15+"; warning: "+w1+", "+w5+", "+w15+"; critical: "+c1+", "+c5+", "+c15
PERFDATA = "nb_cpus="+nb_cpus+";load_1="+l1+";load_5="+l5+";load_15="+l15

print STATUS + "|" + PERFDATA

sys.exit(status[1])

Consommation cpu, mémoire et nombre de process - Solaris

Le script ci-dessous prend comme argument une chaine de caractères qui doit correspondre à un ou plusieurs process qui tournent. Il va alors retourner le nombre de process correspondants trouvés, ainsi que la somme de leur consommation cpu et mémoire. Fonctionne sous Solaris.

#!/bin/ksh
#
# Script de surveillance pour nagios
# Affiche la consommation memoire et cpu des process passes en parametre
# ainsi que le nombre de process trouves


###################### Variables #################################
TEMP_FILE=/tmp/$$-1.tmp
MEM=0
CPU=0
PROCESS_NB=0
STATUS=0

###################### Fonctions #################################

usage() {
    echo "Usage: $0 process(es)-to-analyse"
}

##################################################################
# Programme principal                                            #
##################################################################

if [ "$#" -ne 1 ]
then
    echo "Wrong parameters number."
    usage
    exit 1
fi


PROCESS=$1
ps -edf > $TEMP_FILE
PROCESS_LIST=$(cat "$TEMP_FILE" |grep "$PROCESS"| grep -v "grep $PROCESS" | grep -v "$0 $1" |awk '{print $2}')


if [ "$PROCESS_LIST" == "" ]
then
    echo "Unable to find such running process."
    TEXT="No process"
    STATUS=2
else
    for p in $PROCESS_LIST
    do
        INFO=$(prstat -p $p 1 1 |grep $p|awk '{print $3 " " $9}') # mem=$3, cpu=$9
        MEM_INFO=$(echo "$INFO" | awk '{print $1}')
        CPU_INFO=$(echo "$INFO" | awk '{print $2}')
        MEM="$MEM""+"$(echo "$MEM_INFO" | sed "s/K/\*1024/g" | sed "s/M/\*1024\*1024/g" | sed "s/G/\*1024\*1024\*1024/g")
        CPU="$CPU""+"$(echo "$CPU_INFO" | sed "s/%//g")
        PROCESS_NB=$(($PROCESS_NB+1))
    done
    MEM=$(echo "$MEM"|bc)               # Memoire utilisee en octets
    CPU=$(echo "scale=1;$CPU"|bc)       # Cpu utilise en %

    TEXT="Status Ok"
fi


# On fait le menage
rm -f $TEMP_FILE

echo "${TEXT}|nb=${PROCESS_NB} mem=${MEM} cpu=${CPU}"
exit $STATUS

Exemple :

# ./check_process_ressources ftpd
Status Ok|nb=12 mem=50987008 cpu=0