Lambda alias routing changes don't seem to affect what gets invoked.

Lambda aliases are indeed designed to facilitate safe deployments, rollbacks, and routing of traffic between different versions of your Lambda function. However, the behavior you're seeing indicates a possible misconfiguration. When you have an API Gateway endpoint triggering a Lambda function, it's essential to ensure that the API Gateway is configured to invoke the correct Lambda alias, not the $LATEST version.

Here are the steps to troubleshoot and resolve the issue:

1. API Gateway Configuration:

• Go to the AWS Management Console.
• Navigate to API Gateway.
• Select your API and go to the method (e.g., GET, POST) that triggers the Lambda function.
• In the Integration Request section, check the Lambda function that is being triggered. Ensure it's pointing to the alias (e.g., my-lambda-function:active) and not to the $LATEST`` version (e.g., my-lambda-function).
• If it's not pointing to the correct alias, update the Lambda function ARN to include the alias and save the changes.

2. Deployment:

• After making changes in API Gateway, always remember to **Deploy **the API to the desired stage. This action pushes your configuration changes to the selected environment.

3. Check CloudWatch Logs:

• After updating API Gateway and deploying the changes, invoke the API endpoint and check the CloudWatch logs for your Lambda function. Ensure that the correct version (or alias) is being triggered.

4. Lambda Configuration:

• Ensure that the "active" alias in Lambda is pointing to the correct version (10 in your case). Although you mentioned you checked this, it's always good to double-check.

5. Caching:

• If you have caching enabled on the API Gateway stage, consider clearing the cache after making changes to ensure you're not getting cached responses.

6. CloudFront Distribution (if using API Gateway with Edge Optimized):

• If your API Gateway endpoint is of type "Edge Optimized", it uses a CloudFront distribution behind the scenes. It might take some time for configuration changes to propagate across all edge locations. If you've made the correct changes and the behavior hasn't updated immediately, consider waiting a bit longer.

If you've followed these steps and the issue persists, consider reaching out to AWS Support for further assistance, as there might be more specific configurations or nuances in your setup affecting the behavior.