Create a StatusCheckFailed alarm

create_status_alarm.py

import boto.ec2.cloudwatch
import sys
import os

AWS_KEY = os.environ.get("AWS_ACCESS_KEY_ID")
AWS_SECRET = os.environ.get("AWS_SECRET_ACCESS_KEY")
AWS_REGION = os.environ.get("AWS_EC2_REGION", "us-east-1")
TOPIC = 'YOUR_TOPIC'

def create_status_alarm(instance_id):
    ec2_conn = boto.ec2.connect_to_region(AWS_REGION,
        aws_access_key_id=AWS_KEY,
        aws_secret_access_key=AWS_SECRET)
    cloudwatch_conn = boto.ec2.cloudwatch.connect_to_region(
        AWS_REGION,
        aws_access_key_id=AWS_KEY,
        aws_secret_access_key=AWS_SECRET)

    reservations = ec2_conn.get_all_instances(filters = {'instance-id': instance_id})
    if reservations and reservations[0].instances:
        instance = reservations[0].instances[0]
        instance_name = instance.tags['Name']
    else:
        print "Invalid instance-id!"
        sys.exit(1)
    alarm = boto.ec2.cloudwatch.alarm.MetricAlarm(
        connection = cloudwatch_conn,
        name = instance_name + "-status-alarm",
        metric = 'StatusCheckFailed',
        namespace = 'AWS/EC2',
        statistic = 'Maximum',
        comparison = '>=',
        description = 'status check for %s %s' % (instance_id, instance_name),
        threshold = 1.0,
        period = 60,
        evaluation_periods = 2,
        dimensions = {'InstanceId':instance_id},
        alarm_actions = [TOPIC],
        ok_actions = [TOPIC],
        insufficient_data_actions = [TOPIC])
    cloudwatch_conn.put_metric_alarm(alarm)


if __name__ == '__main__':
    if len(sys.argv) < 2:
        print "Usage: create_status_alarm.py <instanceid>"
        sys.exit(2)
    create_status_alarm(sys.argv[1])

Very nice. Helpful demonstration of something I'm building today. Thanks.