Check AWS RDS cpu usage - A simple example for using Cloudwatch metrics in Naemon/Nagios

check_rds_cpu.sh

#!/bin/bash
# syntax: check_rds_cpu.sh <1:db ident> <2:max delay> <3:warn> <4:crit>
#    e.g: check_sqs_age.sh db1 300 70 85

db_name="${1/*./}"
# max_daly in seconds
max_delay=$2
warn=$3
crit=$4

base_delay=360

msg_append(){
        if [ -z "$msg" ]; then
                msg="$1"
        else
                msg="$msg, $1"
        fi
}

ec_esca(){ [ ${ec:-0} -lt $1 ] && ec=$1; }

pptime(){
        if [ $1 -gt 129600 ]; then
                echo "$(($1/86400))d$(($1%86400/3600))h$(($1%3600/60))m"
        elif [ $1 -gt 5400 ]; then
                echo "$(($1/3600))h$(($1%3600/60))m"
        elif [ $1 -gt 120 ]; then
                echo "$(($1/60))m$(($1%60))s"
        else
                echo "$(($1))s"
        fi
}

get_cw(){
        unset o
        # retries for cloudwatch eventual consistancy (in a monitoring service?!?!?!?)
        for i in `seq 2 10`; do
                o=$(aws cloudwatch get-metric-statistics --namespace AWS/RDS \
                        --metric-name CPUUtilization \
                        --dimensions Name=DBInstanceIdentifier,Value=$db_name \
                        --start-time $((`date +%s`/60*60-60*$i)) --end-time $((`date +%s`/60*60-60*($i-1))) \
                        --period 60 --statistics Maximum --output text | sed 1d)
                [ ! -z "$o" ] && break
        done
        if [ ! -z "$o" ]; then
                read D m t u <<< "$o"
                echo ${m/.*/} $((`date +%s`-`date -d "$t" +%s`))
        else
                echo "NA NA"
        fi
}

read data delay <<< "`get_cw`"

perf_data="data:$data; delay:$delay;"

### check delay
###############

if [ ${delay} -gt $max_delay ]; then
        msg_append "delay too big (`pptime $delay`>`pptime $max_delay`)"
        ec_esca 2
fi

### check data
#######################

if [ "$data" = "NA" ]; then
        msg_append "unable to get cloudwatch data point"
        ec_esca 2
elif [ $data -gt $crit ]; then
        msg_append "Cpu usage $data"
        ec_esca 2
elif [ $data -gt $warn ]; then
        msg_append "Cpu usage $data"
        ec_esca 1
else
        msg_append "Cpu usage $data"
        ec_esca 0
fi

echo "$msg | $perf_data"
#echo "<pre>$info</pre>"
exit $ec

rds.cfg

# input: <rds name> <datapoint max delay> <warn> <crit>
define command{
        command_name    check_rds_cpu
        command_line    $USER1$/check_rds_cpu.sh $HOSTNAME$ $ARG1$ $ARG2$ $ARG3$
        }