How to detect drift in AWS stack- part 2

In the first part, I had explained how to detect drift manually. Now we will see how to automate that.

We are going to write a lambda function which can tell us whether there is any drift present in a given stack. There are boto3 api which helps to get this information. We can schedule this lambda to run at a given schedule or can run it on demand to check for drift.


import boto3
import pprint

client = boto3.client('cloudformation')
pp = pprint.PrettyPrinter(indent=4)

response = client.detect_stack_drift(


{   'ResponseMetadata': {   'HTTPHeaders': {   'content-length': '365',
                                               'content-type': 'text/xml',
                                               'date': 'Mon, 27 Jul 2020 '
                                                       '14:20:45 GMT',
                                               'x-amzn-requestid': '9816363a-f195-4e16-981f-e9d58a943a05'},
                            'HTTPStatusCode': 200,
                            'RequestId': '9816363a-f195-4e16-981f-e9d58a943a05',
                            'RetryAttempts': 0},
    'StackDriftDetectionId': '5c644560-d014-11ea-8be2-027feff0383e'}

Response gives a 'StackDriftDetectionId' which we are going to use to get more information.

Drift status

drift_detection_response = client.describe_stack_drift_detection_status(


and the response -

{'StackId': 'arn:aws:cloudformation:ap-south-1:123456789:stack/my-test-stack/d7d22f50-b70f-11ea-b1ed-0207d27cfd0c',
 'StackDriftDetectionId': '5c644560-d014-11ea-8be2-027feff0383e',
 'StackDriftStatus': 'DRIFTED',
 'DetectionStatus': 'DETECTION_COMPLETE',
 'DriftedStackResourceCount': 1,
 'Timestamp': datetime.datetime(2020, 7, 27, 14, 20, 45, 622000, tzinfo=tzutc()),
 'ResponseMetadata': {'RequestId': '13b04fef-7628-49f9-a4d6-d0fb65651efd',
  'HTTPStatusCode': 200,
  'HTTPHeaders': {'x-amzn-requestid': '13b04fef-7628-49f9-a4d6-d0fb65651efd',
   'content-type': 'text/xml',
   'content-length': '780',
   'date': 'Mon, 27 Jul 2020 14:24:03 GMT'},
  'RetryAttempts': 0}}

StackDriftStatus of this stack is "DRIFTED".  

Drift details

Now to get the details about the drift, we can query the stack for all the resource where drift status is modified.

response = client.describe_stack_resource_drifts(

This response gives us a list of all the drifts in the stack. Each object in the list, among other properties, gives you ExpectedProperties, ActualProperties and PropertyDifferences.

	'ActualProperties': '{"BucketName":"my-fav-collection","VersioningConfiguration":{"Status":"Suspended"}}',
    'ExpectedProperties': '{"BucketName":"my-fav-collection","VersioningConfiguration":{"Status":"Enabled"}}',
    'LogicalResourceId': 'MyS3Bucket',
    'PhysicalResourceId': 'my-fav-collection',
    'PropertyDifferences': [   {   'ActualValue': 'Suspended',
                                   'DifferenceType': 'NOT_EQUAL',
                                   'ExpectedValue': 'Enabled',
                                   'PropertyPath': '/VersioningConfiguration/Status'}],
    'ResourceType': 'AWS::S3::Bucket',
    'StackId': 'arn:aws:cloudformation:ap-south-1:844280065683:stack/my-test-stack/d7d22f50-b70f-11ea-b1ed-0207d27cfd0c',
    'StackResourceDriftStatus': 'MODIFIED',
    'Timestamp': datetime.datetime(2020, 7, 27, 14, 20, 46, 478000, tzinfo=tzutc())

Here, PropertyDifferences gives the propertyPath, its Actual value and ExpectedValue. With this we can identify where the stack is not in sync with cloud formation template.

Complete code

import boto3
client = boto3.client('cloudformation')

def drift_status(stack_name):
        response = client.detect_stack_drift(StackName=stack_name)
        drift_detection_status = client.describe_stack_drift_detection_status(
        if drift_detection_status['StackDriftStatus'] == 'DRIFTED':
            stack_resource_drifts_response = client.describe_stack_resource_drifts(
            return stack_resource_drifts_response['StackResourceDrifts']
        return None

Next, we will schedule this lambda to run regularly and update you about any drift.

Show Comments