AWS SageMaker标注任务Lambda事件为空问题求助

Let's break down why your Lambda is receiving an empty event dictionary and how to fix it—this is usually a configuration or trigger issue rather than a code problem, since you've already set up the necessary permissions.

1. You're Testing the Lambda Manually (Not Triggered by SageMaker) #

The Annotation Consolidation Lambda is only invoked automatically by SageMaker when there are completed annotations to consolidate. If you're using the Lambda console's "Test" feature with a blank or custom test event, you'll get an empty event because SageMaker hasn't passed any data to it.

Fix: #

• Wait until your private workforce has submitted at least one annotation for the labeling job. SageMaker will trigger your Lambda once annotations are ready for consolidation.
• If you want to test locally, use the example event structure you provided to create a test event in the Lambda console.

2. Lambda Resource Policy Doesn't Allow SageMaker to Invoke It #

Even with AmazonSageMakerFullAccess on your Lambda role, you need to explicitly grant SageMaker permission to call your Lambda function. Without this, SageMaker can't send the event data.

Fix: #

Add this resource policy to your Lambda function (under Configuration > Permissions > Resource policy > Add permission):

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Principal": {
        "Service": "sagemaker.amazonaws.com"
      },
      "Action": "lambda:InvokeFunction",
      "Resource": "arn:aws:lambda:us-east-1:YOUR_ACCOUNT_ID:function:YOUR_CUSTOM_LAMBDA_NAME"
    }
  ]
}

Replace placeholders with your actual account ID and Lambda name.

3. Mismatched Regions or Incorrect Lambda ARN #

Double-check that:

• Your labeling job and Lambda function are in the same AWS region (e.g., both in us-east-1). Cross-region invocations won't work here.
• The AnnotationConsolidationLambdaArn in your create_labeling_job code is 100% correct—no typos in the ARN string.

4. Outdated Python Runtime #

Python 3.6 is no longer supported by AWS Lambda (it reached end-of-life in 2021). Using an unsupported runtime can cause compatibility issues where SageMaker can't properly pass event data to your function.

Fix: #

Update your Lambda's runtime to Python 3.8 or later (under Configuration > Runtime settings > Edit).

5. Pre-Human Task Lambda Is Failing #

If your PRE-BoundingBox Lambda is throwing errors, it can halt the entire labeling workflow before annotations reach the consolidation step. SageMaker won't trigger your post-annotation Lambda if the pre-task step fails.

Fix: #

Check the CloudWatch logs for your pre-task Lambda to ensure it's returning the expected response format for SageMaker.

Robust Lambda Code with Error Handling #

To avoid crashes from missing keys and get better debug info, modify your Lambda code to use safe access and error handling:

import json
import boto3
from urllib.parse import urlparse

def get_s3_client(role_arn, kms_key_id):
    sts_client = boto3.client('sts')
    assumed_role = sts_client.assume_role(
        RoleArn=role_arn,
        RoleSessionName='SageMakerConsolidationSession'
    )
    return boto3.client(
        's3',
        aws_access_key_id=assumed_role['Credentials']['AccessKeyId'],
        aws_secret_access_key=assumed_role['Credentials']['SecretAccessKey'],
        aws_session_token=assumed_role['Credentials']['SessionToken']
    )

def consolidate_annotations(labeling_job_arn, payload, label_attr, s3_client):
    # Add your consolidation logic here
    # Example: Fetch annotations from payload['s3Uri'], process, return results
    return {"consolidatedLabel": "example_result"}

def lambda_handler(event, context):
    print("Received event:", json.dumps(event, indent=2))
    
    # Handle empty event case
    if not event:
        print("Warning: Empty event received. This Lambda should be triggered by SageMaker.")
        return {
            "statusCode": 400,
            "body": json.dumps("Empty event - Lambda must be invoked by SageMaker Annotation Consolidation")
        }
    
    try:
        # Safely extract event fields
        labeling_job_arn = event.get("labelingJobArn")
        label_attr = event.get("labelAttributeName")
        payload = event.get("payload")
        role_arn = event.get("roleArn")
        
        # Validate critical fields
        if not all([labeling_job_arn, label_attr, payload, role_arn]) or "s3Uri" not in payload:
            raise ValueError("Missing required fields in event")
        
        parsed_s3_uri = urlparse(payload["s3Uri"])
        print("Parsed annotation S3 URI:", parsed_s3_uri)
        
        label_categories = event.get("labelCategories", "None provided")
        print("Label categories:", label_categories)
        
        # Initialize S3 client and run consolidation
        s3_client = get_s3_client(role_arn, event.get("kmsKeyId"))
        consolidation_result = consolidate_annotations(labeling_job_arn, payload, label_attr, s3_client)
        
        return {
            "statusCode": 200,
            "body": json.dumps(consolidation_result)
        }
    
    except Exception as e:
        print(f"Consolidation failed: {str(e)}")
        return {
            "statusCode": 500,
            "body": json.dumps(f"Error during annotation consolidation: {str(e)}")
        }

Final Debug Step #

After applying these fixes, check your Lambda's CloudWatch logs (under Monitor > Logs > Log groups) once annotations are submitted. You'll see the full event payload from SageMaker if everything is configured correctly.

内容的提问来源于stack exchange，提问作者José R. Sánchez