SageMaker Ground Truth Recipe

Pre-labeling and Post-labeling lambdas for custom labeling jobs. If you want to just use these lambdas, you can directly import "aws-sagemaker-ground-truth-recipe" in your AWS console. You can also modify code in your AWS console.

If you want to use Serveless Application Model toolkit to modify and develop the code further, keep reading.

Requirements

• AWS CLI already configured with Administrator permission
• Python 3 installed
• Docker installed

Setup process

Local development

Invoking function locally using a local sample payload

Testing GtRecipePreHumanTaskFunction Lambda

sam local invoke GtRecipePreHumanTaskFunction --event resources/pre_human_task_test_event.json 

Testing GtRecipeAnnotationConsolidationFunction Lambda

Following function will fail due to permission issue. Modify annotation_consolidation_test_event.json before trying. Change "roleArn" to your ARN Change "s3Uri" to point to a JSON file. A sample S3 file is located here

sam local invoke GtRecipeAnnotationConsolidationFunction --event resources/annotation_consolidation_test_event.json 

Note : You can also test GtRecipeAnnotationConsolidationFunction in AWS UI Console after deployment

Packaging and deployment

AWS Lambda Python runtime requires a flat folder with all dependencies including the application. SAM will use CodeUri property to know where to look up for both application and dependencies:

...
    GtRecipePreHumanTaskFunction:
        Type: AWS::Serverless::Function
        Properties:
            CodeUri: aws_sagemaker_ground_truth_sample_lambda/
            ...

Firstly, we need a S3 bucket where we can upload our Lambda functions packaged as ZIP before we deploy anything - If you don't have a S3 bucket to store code artifacts then this is a good time to create one:

aws s3 mb s3://BUCKET_NAME

Next, run the following command to package our Lambda function to S3:

sam package \
    --output-template-file packaged.yaml \
    --s3-bucket REPLACE_THIS_WITH_YOUR_S3_BUCKET_NAME

Next, the following command will create a Cloudformation Stack and deploy your SAM resources.

sam deploy \
    --template-file packaged.yaml \
    --stack-name gt-recipe \
    --capabilities CAPABILITY_IAM

See Serverless Application Model (SAM) HOWTO Guide for more details in how to get started.

After deployment is complete you can run the following command to retrieve the API Gateway Endpoint URL:

aws cloudformation describe-stacks \
    --stack-name gt-recipe 

Fetch, tail, and filter Lambda function logs

To simplify troubleshooting, SAM CLI has a command called sam logs. sam logs lets you fetch logs generated by your Lambda function from the command line. In addition to printing the logs on the terminal, this command has several nifty features to help you quickly find the bug.

NOTE: This command works for all AWS Lambda functions; not just the ones you deploy using SAM.

sam logs -n GtRecipePreHumanTaskFunction --stack-name gt-recipe --tail

You can find more information and examples about filtering Lambda function logs in the SAM CLI Documentation.

Testing

Next, we install test dependencies and we run pytest against our tests folder to run our initial unit tests:

pip install pytest pytest-mock --user
python -m pytest tests/ -v

Cleanup

In order to delete our Serverless Application recently deployed you can use the following AWS CLI Command:

aws cloudformation delete-stack --stack-name gt-recipe

Bringing to the next level

Here are a few things you can try to get more acquainted with building serverless applications using SAM:

Learn how SAM Build can help you with dependencies

• Build the project with sam build --use-container
• Invoke with sam local invoke GtRecipePreHumanTaskFunction --event resources/annotation_consolidation_test_event.json
• Update tests

Step-through debugging

• Enable step-through debugging docs for supported runtimes

Next, you can use AWS Serverless Application Repository to deploy ready to use Apps that go beyond hello world samples and learn how authors developed their applications: AWS Serverless Application Repository main page

Appendix

Building the project

AWS Lambda requires a flat folder with the application as well as its dependencies in deployment package. When you make changes to your source code or dependency manifest, run the following command to build your project local testing and deployment:

sam build

If your dependencies contain native modules that need to be compiled specifically for the operating system running on AWS Lambda, use this command to build inside a Lambda-like Docker container instead:

sam build --use-container

By default, this command writes built artifacts to .aws-sam/build folder.

SAM and AWS CLI commands

All commands used throughout this document

# Invoke function locally with event.json as an input
sam local invoke GtRecipePreHumanTaskFunction --event resources/pre_human_task_test_event.json

# Create S3 bucket
aws s3 mb s3://BUCKET_NAME

# Package Lambda function defined locally and upload to S3 as an artifact
sam package \
    --output-template-file packaged.yaml \
    --s3-bucket REPLACE_THIS_WITH_YOUR_S3_BUCKET_NAME

# Deploy SAM template as a CloudFormation stack
sam deploy \
    --template-file packaged.yaml \
    --stack-name gt-recipe \
    --capabilities CAPABILITY_IAM

# Describe Output section of CloudFormation stack previously created
aws cloudformation describe-stacks \
    --stack-name gt-recipe
    
# Tail Lambda function Logs using Logical name defined in SAM Template
sam logs -n GtRecipePreHumanTaskFunction --stack-name gt-recipe --tail