Production ready serverless

Coding along while following the Production ready serverless course.

Overview

• CDK deployment: cdk/app.py

• REST API:

  • lambda functions exposed behing a REST API Gateway, handled with lambda power tools and pydantic data models
  • endpoints:
    • /restaurants:
      • internal API listing all restaurants from DynamoDB
      • protected with IAM
    • /restaurants/search:
      • search by attribute in DynamoDB
      • protected with Cognito
    • /orders:
      • used to post new orders
      • protected with Cognito
      • asynchronous processing: event is pushed to EventBridge, then processed by a lambda
    • / :
      • public HTML page
      • queries data from /restaurants on server side, signing requests with sigv4
      • allows users to register or sign in to the Cognito user pool, using SRP
      • uses the Cognito JWT token to send requests from the browser to /restaurants/search

• Lambda Power Tools is used for REST request handling, idempotency, event sourced event parsing and logging

• database: DynamoDB

• Tests:

  • BDD style integration tests and end-to-end tests using pytest-bdd
  • Event-driven cases are validated by creating temporary SQS test queues to capture the events during the tests

Dev setup

python3.12 -m venv .venv-box
source .venv-box/bin/activate
pip install -r cdk/requirements.txt

Shared SSM parameters

In this project, SSM is used for the configuration shared across different deployments, following the convention:

/{SERVICE_NAME}/shared_context/{MATURITY_LEVEL}/..."

where:

• SERVICE_NAME is the name of the service, e.g. production-ready-serverless
• MATURITY_LEVEL is linked to the release life cycle, e.g. dev, test, acc, prod

Note that the FEATURE_NAME, used in the deployed stack name, is not part of the SSM parameter path.

Those parameters are expected to be created before the deployment and their value is shared across all deployments having the same maturity level (each with a different feature name). This allows to share contextual information, like the URL of 3rd API,...

For dynamic configuration that should not be shared across deployments, e.g. feature flags, we could use a convention similar to the following instead:

# (not currently implemented in this demo project)
/{SERVICE_NAME}/{FEATURE_NAME}/..."

How to use

The FEATURE_NAME env variable is used to prefix the stack and API names.

Build and deploy:

MATURITY_LEVEL=dev \
FEATURE_NAME=feature-foo \
  cdk deploy \
  --all

Deployment delta:

MATURITY_LEVEL=dev \
FEATURE_NAME=feature-foo \
  cdk diff

Seed the DB:

MATURITY_LEVEL=dev \
FEATURE_NAME=feature-foo \
  python seed/seed_restaurants.py

See stack output for the app URL, then open it in a browser.

Tests

Both the integration and end-to-end tests require resources to already be present in the AWS account.

The database should also have been seeded with the script above to match exactly the test expectations (TODO: automate this step).

Executing the tests requires to be authenticated with AWS.

Create a virtualenv with the required dependencies

pip install -r functions/get_index/requirements.txt
pip install -r tests/requirements.txt

Integration tests

Integration tests are invoking the lambdas handlers directly. The MATURITY_LEVEL and FEATURE_NAME environment variables are used to retrieve AWS resources details.

MATURITY_LEVEL=dev \
FEATURE_NAME=feature-foo \
PYTHONPATH=src/functions/get_index:src/functions/get_restaurants:src/functions/search_restaurants \
  pytest tests/integration \
  -s \
  -v \
  --gherkin-terminal-reporter -v

End-to-end tests

Integration tests are invoking the REST endpoints exposed via the API-gateway.

Temporary Cognito users are created and deleted during the tests.

PYTHONPATH=src/functions/place_order \
FEATURE_NAME=feature-foo \
  pytest tests/end-to-end \
  -s \
  -v \
  --gherkin-terminal-reporter -v

Add -k to run a specific test file, e.g.:

PYTHONPATH=src/functions/place_order \
FEATURE_NAME=feature-foo \
  pytest tests/end-to-end \
  -s \
  -v \
  --gherkin-terminal-reporter -v \
  -k test_orders_scenarios.py