Using Nagios to monitor OST capacity

Show TOCHide TOC

1 How to use this plugin

• This plugin relies on check_nrpe, so you have to have such capabilities on the local host
• This plugin must run on a lustre client
• Keep in mind that many versions of check_nrpe have a 1024 character limit on output
• usage currently requires -f, c, and w arguments, e.g.: ./check_osts.sh -f naaschpc -c 90 -w 89

2 Plugin script

#!/usr/bin/env bash

# Nagios Exit Codes
OK=0
WARNING=1
CRITICAL=2
UNKNOWN=3

usage()
{
cat <<EOF

Check the fullness of lustre OSTs.

     Options:
        -f         Specify filesystem 
        -c         Critical threshold as an int (0-100)
        -w         Warning threshold as an int (0-100)

Usage: $0 -f tlustre -c 10 -w 6
EOF
}

if [ $# -lt 6 ];
then
   usage
   exit 1
fi

# Define now to prevent expected number errors
FS=mylustre
CRIT=0
WARN=0

while getopts "hc:f:w:" OPTION
do
     case $OPTION in
         h)
        usage
             ;;
         c)
        CRIT="$OPTARG"
             ;;
    f)
         FS="$OPTARG"
        ;;
    w)
        WARN="$OPTARG"
        ;;
         \?)
             exit 1
             ;;
     esac
done

#use the fullest OST to determine CRIT or WARN 
STATUS=$(lfs df | grep ${FS}-OST |awk '{sub(/_UUID/,"")}; 1' |awk  '{sub(/%/,"")}; 1 {printf "%s\n", $5}' | sort -gr | head -1)

#gather the data on all OSTs
ALLSORT=$(lfs df | grep ${FS}-OST |sed -e "s/^$FS-//" -e "s/_UUID//" |awk  '{sub(/%/,"")}; 1 {printf "%s %s\n", $1, $5}' | awk '{printf "%s=%s% ", $1, $2}')

# you never need to do the in-between CRIT and WARN logic because of the order of these if statements....
if [ $STATUS -gt $CRIT ]; then
       echo "At least one OST is at ${STATUS}% capacity! | $ALLSORT" 
       exit $CRITICAL
elif [ $STATUS -gt $WARN ]; then
       echo "At least one OST is at ${STATUS}% capacity! | $ALLSORT"
       exit $WARNING
else
       echo "All OSTs are below warning capacity. | $ALLSORT" 
       exit $OK
fi