cloudwatch-create-alarms.sh

#!/usr/bin/env bash
# This script creates AWS CloudWatch alarms based on standard metrics and user input to setup alarms for each environment
# Requires AWS CLI Setup and you must setup your ALARMACTION

ALARMACTION="arn:aws:sns:us-east-1:YOURACCOUNTNUMBER:YOURSNSALERTNAME"

# Functions

# Check Command
function check_command {
  type -P $1 &>/dev/null || fail "Unable to find $1, please install it and run this script again."
}

# Completed
function completed(){
  echo
  HorizontalRule
  tput setaf 2; echo "Completed!" && tput sgr0
  HorizontalRule
  echo
}

# Fail
function fail(){
  tput setaf 1; echo "Failure: $*" && tput sgr0
  exit 1
}

# Horizontal Rule
function HorizontalRule(){
  echo "============================================================"
}


# Verify AWS CLI Credentials are setup
# http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html
if ! grep -q aws_access_key_id ~/.aws/config; then
  if ! grep -q aws_access_key_id ~/.aws/credentials; then
    fail "AWS config not found or CLI not installed. Please run \"aws configure\"."
  fi
fi

# Check for AWS CLI profile argument passed into the script
# http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html#cli-multiple-profiles
if [ $# -eq 0 ]; then
  scriptname=`basename "$0"`
  echo "Usage: ./$scriptname profile"
  echo "Where profile is the AWS CLI profile name"
  echo "Using default profile"
  echo
  profile=default
else
  profile=$1
fi

# Verify ALARMACTION is setup with some alert mechanism
if [[ -z $ALARMACTION ]] || [[ "$ALARMACTION" == "arn:aws:sns:us-east-1:YOURACCOUNTNUMBER:YOURSNSALERTNAME" ]]; then
  echo "Alarm Action SNS Topic ARN?"
  echo "Example: arn:aws:sns:us-east-1:YOURACCOUNTNUMBER:YOURSNSALERTNAME"
  read -r ALARMACTION
  if [[ -z $ALARMACTION ]]; then
    fail "Alarm Action must be configured."
  fi
fi

HorizontalRule
echo "Create CloudWatch Alarms"
HorizontalRule
echo

read -r -p "Client Name? " CLIENT
if [[ -z $CLIENT ]]; then
  fail "Invalid Client Name!"
fi
HorizontalRule
read -r -p "How Many Servers Total? " SERVERNUM
if [[ $SERVERNUM > 0 ]] && echo "$SERVERNUM" | egrep -q '^[0-9]+$'; then
  read -r -p "Loadbalanced Environment? (y/n) " LOADBALANCED

  # If Loadbalanced Environment
  if [[ $LOADBALANCED =~ ^([yY][eE][sS]|[yY])$ ]]; then
    read -r -p "Load Balancer ID? " LBID
    if [[ -z $LBID ]]; then
      fail "Invalid Load Balancer ID!"
    fi

    # Load Balancer Unhealthy Host Check
    aws cloudwatch put-metric-alarm --alarm-name "$CLIENT Unhealthy Host Check" --alarm-description "$CLIENT Load Balancer Unhealthy Host Detected" --metric-name "UnHealthyHostCount" --namespace "AWS/ELB" --statistic "Sum" --period 60 --threshold 0 --comparison-operator "GreaterThanThreshold" --dimensions Name=LoadBalancerName,Value=$LBID --evaluation-periods 3 --alarm-actions "$ALARMACTION" --profile $profile
    HorizontalRule
    echo "Load Balancer Unhealthy Host Alarm Set"
    HorizontalRule
    # Load Balancer High Latency Check
    aws cloudwatch put-metric-alarm --alarm-name "$CLIENT LB High Latency" --alarm-description "$CLIENT Load Balancer High Latency" --metric-name "Latency" --namespace "AWS/ELB" --statistic "Average" --period 60 --threshold 15 --comparison-operator "GreaterThanThreshold" --dimensions Name=LoadBalancerName,Value=$LBID --evaluation-periods 2 --alarm-actions "$ALARMACTION" --profile $profile
    HorizontalRule
    echo "Load Balancer High Latency Alarm Set"
    HorizontalRule
  fi

  # Begin loop to create server alarms
  START=1
  for (( COUNT=$START; COUNT<=$SERVERNUM; COUNT++ )) do
    echo "Server #"$COUNT
    read -r -p "Server Environment? (Dev/Staging/Production) " ENVIRONMENT
    if [[ -z $ENVIRONMENT ]]; then
      fail "Invalid Server Environment!"
    fi
    read -r -p "Server Name? (Web01, Web02) " SERVERNAME
    # Avoid "Dev Dev" situation
    if [[ "$ENVIRONMENT" == "$SERVERNAME" ]]; then
      SERVERNAME=""
    fi
    read -r -p "Instance ID? (i-xxxxxxxx or i-xxxxxxxxxxxxxxxxx) " INSTANCEID
    if [[ "$INSTANCEID" =~ ^([i]-........)|([i]-.................)$ ]]; then

        # CPU Check
        aws cloudwatch put-metric-alarm --alarm-name "$CLIENT $ENVIRONMENT $SERVERNAME CPU Check" --alarm-description "$CLIENT $ENVIRONMENT $SERVERNAME CPU usage >90% for 5 minutes" --namespace "AWS/EC2" --dimensions Name=InstanceId,Value=$INSTANCEID --metric-name "CPUUtilization" --statistic "Average" --comparison-operator "GreaterThanThreshold" --unit "Percent" --period 60 --threshold 90 --evaluation-periods 5 --alarm-actions "$ALARMACTION" --profile $profile
        HorizontalRule
        echo $CLIENT $ENVIRONMENT $SERVERNAME "CPU Check Alarm Set"
        HorizontalRule

        # Status Check
        aws cloudwatch put-metric-alarm --alarm-name "$CLIENT $ENVIRONMENT $SERVERNAME Status Check" --alarm-description "$CLIENT $ENVIRONMENT $SERVERNAME Status Check Failed for 5 minutes" --namespace "AWS/EC2" --dimensions Name=InstanceId,Value=$INSTANCEID --metric-name "StatusCheckFailed" --statistic "Maximum" --comparison-operator "GreaterThanThreshold" --unit "Count" --period 60 --threshold 0 --evaluation-periods 5 --alarm-actions "$ALARMACTION" --profile $profile
        HorizontalRule
        echo $CLIENT $ENVIRONMENT $SERVERNAME "Status Check Alarm Set"
        HorizontalRule
    else
      fail "Invalid Instance ID!"
    fi
  done
else
  if [[ $SERVERNUM == 0 ]]; then
    echo "Skipping Server Alarms..."
  else
    tput setaf 1; echo "Invalid Number of Servers!" && tput sgr0
  fi
fi

read -r -p "Setup Database Alarms? (y/n) " SETUPDB
  # If Database Alarms
  if [[ $SETUPDB =~ ^([yY][eE][sS]|[yY])$ ]]; then
    HorizontalRule
    read -r -p "How Many Database Hosts Total? " DBNUM
    if [[ $DBNUM > 0 ]] && echo "$DBNUM" | egrep '^[0-9]+$' >/dev/null 2>&1; then

    # Begin loop to create database alarms
      START=1
      for (( COUNT=$START; COUNT<=$DBNUM; COUNT++ )) do
        echo "DB #"$COUNT
        read -r -p "Database Environment? (Dev/Staging/Production) " ENVIRONMENT
        if [[ -z $ENVIRONMENT ]]; then
          fail "Invalid Database Environment!"
        fi
        # # Avoid "Beta Beta" situation
        # if [[ $ENVIRONMENT == "Beta" ]]; then
        #   SERVERNAME=""
        # else
        #   echo -n "DB Name? (Web01, Web02) "
        #   read SERVERNAME
        # fi
        read -r -p "DB Instance ID? " DBID
        if [[ -z $DBID ]]; then
          fail "Invalid Database Instance ID!"
        fi

        # Database CPU Check
        aws cloudwatch put-metric-alarm --alarm-name "$CLIENT $ENVIRONMENT DB CPU Check" --alarm-description "$CLIENT $ENVIRONMENT Database CPU usage >90% for 5 minutes" --metric-name "CPUUtilization" --namespace "AWS/RDS" --statistic "Average" --unit "Percent" --period 60 --threshold 90 --comparison-operator "GreaterThanThreshold" --dimensions Name=DBInstanceIdentifier,Value=$DBID --evaluation-periods 5 --alarm-actions "$ALARMACTION" --profile $profile
        HorizontalRule
        echo $CLIENT $ENVIRONMENT "Database CPU Check Alarm Set"
        HorizontalRule

        # Database Memory Usage Check
        aws cloudwatch put-metric-alarm --alarm-name "$CLIENT $ENVIRONMENT DB Mem Check" --alarm-description "$CLIENT $ENVIRONMENT Database Freeable Memory < 200 MB for 5 minutes" --metric-name "FreeableMemory" --namespace "AWS/RDS" --statistic "Average" --unit "Bytes" --period 60 --threshold "200000000" --comparison-operator "LessThanThreshold" --dimensions Name=DBInstanceIdentifier,Value=$DBID --evaluation-periods 5 --alarm-actions "$ALARMACTION" --profile $profile
        HorizontalRule
        echo $CLIENT $ENVIRONMENT "Database Memory Usage Alarm Set"
        HorizontalRule

        # Database Available Storage Space Check
        aws cloudwatch put-metric-alarm --alarm-name "$CLIENT $ENVIRONMENT DB Storage Check" --alarm-description "$CLIENT $ENVIRONMENT Database Available Storage Space < 200 MB" --metric-name "FreeStorageSpace" --namespace "AWS/RDS" --statistic "Average" --unit "Bytes" --period 60 --threshold "200000000" --comparison-operator "LessThanThreshold" --dimensions Name=DBInstanceIdentifier,Value=$DBID --evaluation-periods 1 --alarm-actions "$ALARMACTION" --profile $profile
        HorizontalRule
        echo $CLIENT $ENVIRONMENT "Database Available Storage Space Alarm Set"
        HorizontalRule
      done
    else
      if [[ $DBNUM == 0 ]]; then
        echo "Skipping Database Alarms..."
      else
        tput setaf 1; echo "Invalid Number of Databases!" && tput sgr0
      fi
    fi
  else
    echo "Exiting"
  fi
completed