Why is my CloudWatch alarm not being applied to the EC2 instances? - aws-lambda

Question

I have python code in a Lambda function to apply a CloudWatch alarm to EC2 instances.
The CloudWatch alarm is to reboot them if they are non-responsive for 10 minutes. This alarm is easy to make on a per EC2 instance basis, but that's a lot of manual work, we have many servers.
I've set up a CloudWatch rule that triggers my Lambda function when an EC2 instance enters the ''running'' state after a reboot, or after a new EC2 instance is launched and gets to ''running''.
I have tried specifying a specific server in my code, and that works. However, what I want is a piece of code that applies it to servers as they are rebooted; so at to cover them all as maintenance windows come around and they all get rebooted.
from collections import defaultdict
import boto3
ec2_sns = 'SNS-Topic:'
ec2_rec ="arn:aws:automate:eu-central-1:ec2:recover"
def lambda_handler(event, context):
ec2 = boto3.resource('ec2')
cw = boto3.client('cloudwatch')
ec2info = defaultdict()
running_instances = ec2.instances.filter(Filters=[{'Name': 'tag-
key','Values': ['cloudwatch'],}])
for instance in running_instances:
for tag in instance.tags:
if 'Name'in tag['Key']:
name = tag['Value']
ec2info[instance.id] = {'Name':
name,'InstanceId':instance.instance_id,}
attributes = ['Name','InstanceId']
for instance_id, instance in ec2info.items():
instanceid =instance["InstanceId"]
nameinsta = instance["Name"]
print(instanceid,nameinsta )
#Create StatusCheckFailed Alamrs
cw.put_metric_alarm(
AlarmName = ('InstanceId') +
"_System_Unresponsive_(Created by Lambda)",
AlarmDescription='System_unresponsive for 10
minutes',
ActionsEnabled=True,
OKActions=[
'No data',
],
AlarmActions=[
'arn:aws:lambda:eu-central
1:788677770941:function:System_unresponsive:reboot',
],
InsufficientDataActions=[
'Insuficient data',
],
MetricName='StatusCheckFailed',
Namespace='AWS/EC2',
Statistic='Average',
Dimensions=[ {'Name': "InstanceId",'Value':
instanceid},],
Period=300,
Unit='Seconds',
EvaluationPeriods=2,
DatapointsToAlarm=2,
Threshold=1,
ComparisonOperator='LessThanOrEqualToThreshold')
I expect that the code applies the specified CloudWatch alarm to servers as they are rebooted, but it doesn't.
When you test it all you get is ''null'' as a result.

Answer 1

You can use CloudTrail to get insights into the API calls that AWS is doing to start the instances and catch just those specific events with CloudWatch Events.
Once you catch the right events and send them to a lambda, the lambda will receive the instance ID in the event information. You could use that information to create/update the alarms just for the instance contained in the event. You could use print(json.dumps(event)) inside the function to inspect the event contents in CloudWatch Logs.