如何使用 Boto3 一次性创建多个 cloudwatch 警报

Question

I want to create ~267 Cloudwatch alarms, manual process is so pathetic, can someone guide me to use Boto3 script so that I can set up all alarms in a one shot.

import boto3

# Create CloudWatch client
cloudwatch = boto3.client('cloudwatch')

# Create alarm
cloudwatch.put_metric_alarm(
    AlarmName='Web_Server_CPU_Utilization',
    ComparisonOperator='GreaterThanThreshold',
    EvaluationPeriods=1,
    MetricName='CPUUtilization',
    Namespace='AWS/EC2',
    Period=60,
    Statistic='Average',
    Threshold=70.0,
    ActionsEnabled=False,
    AlarmDescription='Alarm when server CPU exceeds 70%',
    Dimensions=[
        {
          'Name': 'InstanceId',
          'Value': 'i-xxxxxxxxxx'
        },
    ],
    Unit='Seconds'
)

Answer 1

Assuming you want to add a CloudWatch alarm for different EC2 instances, you can simply put the instance IDs in a list and iterate over that list to create the alarms. That'd look like:

import boto3

cloudwatch = boto3.client('cloudwatch')

ec2_instances = [
    'i-xxxxxxxxx1',
    'i-xxxxxxxxx2',
    'i-xxxxxxxxx3'
]

for ec2_instance in ec2_instances:
    cloudwatch.put_metric_alarm(
        AlarmName='Web_Server_CPU_Utilization_%s' % ec2_instance,
        ComparisonOperator='GreaterThanThreshold',
        EvaluationPeriods=1,
        MetricName='CPUUtilization',
        Namespace='AWS/EC2',
        Period=60,
        Statistic='Average',
        Threshold=70.0,
        ActionsEnabled=False,
        AlarmDescription='Alarm when server CPU exceeds 70%',
        Dimensions=[
            {
              'Name': 'InstanceId',
              'Value': ec2_instance
            },
        ],
        Unit='Seconds'
    )

Answer 2

Here is a simple script I use to set up CloudWatch alarms on my running EC2 instances. The aim is to reboot my EC2 instances if StatusCheckFailed_Instance is True.

In case you are getting the "Insufficient Data" message as well, its worthwhile creating the same alarm on the EC2 console and then making sure your put_metric_alarm call matches the source/CloudFormation JSON.

AWS seems to be really fussy about the JSON. Once I matched the EC2 console's JSON exactly it worked like a charm.

Hope this helps someone.

import boto3

# Specify your region here
region = "ap-northeast-1"

ec2_client = boto3.client("ec2", region_name=region)
cloudwatch = boto3.client('cloudwatch')

# Get running EC2 instances
reservations = ec2_client.describe_instances(Filters=[
    {
        "Name": "instance-state-name",
        "Values": ["running"],
    }
]).get("Reservations")

# Set up an alarm for each instance
for reservation in reservations:
    for instance in reservation["Instances"]:
        instance_id = instance['InstanceId']

        cloudwatch.put_metric_alarm(
            AlarmName=f'Status_Check_{instance_id}',
            AlarmDescription=f'Alarm when status check fails on {instance_id}',
            ActionsEnabled=True,
            OKActions=[],
            AlarmActions=[
                f"arn:aws:automate:{region}:ec2:reboot"
            ],
            InsufficientDataActions=[],
            MetricName='StatusCheckFailed_Instance',
            Namespace='AWS/EC2',
            Statistic='Maximum',
            Dimensions=[
                {
                  'Name': 'InstanceId',
                  'Value': instance_id
                },
            ],
            Period=60,
            EvaluationPeriods=2,
            DatapointsToAlarm=2,
            Threshold=0.99,
            ComparisonOperator='GreaterThanOrEqualToThreshold'
        )