Hello OpenAPI

This repo contains boilerplate code for building AWS Lambda functions using Docker. See notes/decisions.md for stack decisions.

Sample containers are provided:

• hello-deno - simplest example lambda function using Deno
• hello-openapi - opinionated OpenAPI stack

IDE Note

In VSCode, import module resolution can be handled by the denoland.vscode-deno. They suggest keeping it disabled globally and only enabling it per workspace.

Build Docker Image

Build and test the docker image locally.

export CONTAINER_NAME=hello-openapi
docker build -t $CONTAINER_NAME .

(Optional) Run the container locally.

docker run -p 8000:8000 $CONTAINER_NAME:latest
curl http://localhost:8000
docker stop $(docker ps -q)

Deployment

Overview:

1. Create the ECR repository.
2. Push the container image to the ECR repository.
3. Deploy the lambda function.

Create the ECR repository

You'll first need to authenticate with AWS.

export AWS_REGION=us-west-1
export AWS_PROFILE=playground
aws configure sso --profile $AWS_PROFILE

With Terraform:

cd terraform
terraform init
terraform apply -target=aws_ecr_repository.hello_docker
export REPOSITORY_URI=$(terraform output -raw ecr_repository_url)
aws ecr get-login-password --region $AWS_REGION --profile $AWS_PROFILE | docker login --username AWS --password-stdin $REPOSITORY_URI
docker tag $CONTAINER_NAME:latest $REPOSITORY_URI:latest
docker push $REPOSITORY_URI:latest

With aws-sam-cli:

cd sam
sam build
sam deploy --guided --profile playground --region us-west-1 --parameter-overrides DeployECROnly=true
export REPOSITORY_URI=$(aws ecr describe-repositories --repository-names $CONTAINER_NAME --region $AWS_REGION --profile $AWS_PROFILE | jq -r '.repositories[0].repositoryUri')
aws ecr get-login-password --region $AWS_REGION --profile $AWS_PROFILE | docker login --username AWS --password-stdin $REPOSITORY_URI
docker tag $CONTAINER_NAME:latest $REPOSITORY_URI:latest
docker push $REPOSITORY_URI:latest

In either case, we must build the ECR repository before deploying the lambda function. For sam, the DeployECROnly parameter is used to control this behavior for sam. You can set it to false in your samconfig.toml after the first deploy. Thereafter you can use --no-confirm-changeset instead of --guided. For terraform, simply omit the -target parameter to build the full stack.

Deploy the lambda function

With Terraform:

terraform apply

With aws-sam-cli:

sam deploy --guided --profile playground --region us-west-1 --parameter-overrides DeployECROnly=false --resolve-image-repos

Tips

Get the function url for the lambda function

export FUNCTION_NAME=
aws lambda get-function-url-config --function-name $FUNCTION_NAME --query 'FunctionUrl' --output text

Check the image uri for a lambda function

aws lambda get-function --function-name $FUNCTION_NAME --query 'Code.ImageUri' --output text

Review the images in an ECR repository

export REPOSITORY_NAME=hello-world
aws ecr describe-images --repository-name $REPOSITORY_NAME --region $AWS_REGION --profile $AWS_PROFILE

Logs

With Terraform:

export FUNCTION_NAME=hello_docker
aws logs tail /aws/lambda/$FUNCTION_NAME --since 1h --profile $AWS_PROFILE --region $AWS_REGION
aws logs tail /aws/lambda/$FUNCTION_NAME --follow --profile $AWS_PROFILE --region $AWS_REGION

With aws-sam-cli:

sam logs -n HelloWorldFunction --profile playground --tail