Serverless image resizer

Key	Value
Environment
Services	S3, SSM, Lambda, SNS, SES
Integrations	AWS SDK, AWS CLI, GitHub actions, pytest
Categories	Serverless, S3 notifications, S3 website, Lambda function URLs, LocalStack developer endpoints, JavaScript, Python
Level	Intermediate

Introduction

This is an app to resize images uploaded to S3 in a serverless way. A simple web fronted using HTML and JavaScript provides a way for users to upload images that are resized and listed. We use a Lambda to generate S3 pre-signed URLs so the upload form can upload directly to S3 rather than going through the Lambda. S3 bucket notifications are used to trigger a Python Lambda that runs image resizing. Another Lambda is used to list all uploaded and resized images, and provide pre-signed URLs for the browser to display them. We also demonstrate how Lambda failures can submit to SNS, which can then trigger an SES email. Using the LocalStack internal /_localstack/aws/ses endpoint, we can run end-to-end integration tests to verify that emails have been sent correctly.

Here's a short summary of AWS service features we use:

• S3 bucket notifications to trigger a Lambda
• S3 pre-signed POST
• S3 website
• Lambda function URLs
• Lambda SNS on failure destination
• SNS to SES Subscriptions
• SES LocalStack testing endpoint

Here's the web application in action:

Peek.2023-04-01.23-13.mp4

Moreover, the repo includes a GitHub actions workflow to demonstrate how to run end-to-end tests of your AWS apps using LocalStack in CI. The GitHub workflow runs a set of integration tests using pytest.

Architecture overview

Prerequisites

Dev environment

Make sure you use the same version as the Python Lambdas to make Pillow work. If you use pyenv, then first install and activate Python 3.11:

pyenv install 3.11.6
pyenv global 3.11.6

% python --version
Python 3.11.6

Create a virtualenv and install all the development dependencies there:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements-dev.txt

Instructions

You can set up and deploy the sample application on LocalStack by executing the commands in our Makefile. First, create a .env file using the provided .env.example file as a template, and include your LocalStack token in it. Then, run make start to initiate LocalStack on your machine.

Next, execute make install to install needed dependencies.

After that, launch make terraform-setup to provision the infrastructure on LocalStack using Terraform CLI and its scripts. Alternatively, run make awslocal-setup to set up the infrastructure using awslocal, a wrapper for the AWS CLI.

If you prefer, you can also follow these step-by-step instructions for a manual deployment.

LocalStack

Start LocalStack Pro with Auth Token:

LOCALSTACK_AUTH_TOKEN=... localstack start (-d)

Terraform

To create the infrastructure using Terraform, run the following commands:

cd deployment/terraform
tflocal init
tflocal apply --auto-approve

We are using the tflocal wrapper to configure the local service endpoints, and send the API requests to LocalStack, instead of AWS.

AWS CLI

You can execute the following commands to set up the infrastructure using awslocal. All the commands are also available in the deployment/awslocal/deploy.sh script.

Create the buckets

The names are completely configurable via SSM:

awslocal s3 mb s3://localstack-thumbnails-app-images
awslocal s3 mb s3://localstack-thumbnails-app-resized

Put the bucket names into the parameter store

awslocal ssm put-parameter --name /localstack-thumbnail-app/buckets/images --type "String" --value "localstack-thumbnails-app-images"
awslocal ssm put-parameter --name /localstack-thumbnail-app/buckets/resized --type "String" --value "localstack-thumbnails-app-resized"

Create the DLQ Topic for failed lambda invokes

awslocal sns create-topic --name failed-resize-topic

Subscribe an email address to it (to alert us immediately if an image resize fails!).

awslocal sns subscribe \
    --topic-arn arn:aws:sns:us-east-1:000000000000:failed-resize-topic \
    --protocol email \
    --notification-endpoint [email protected]

Create the lambdas

S3 pre-signed POST URL generator

This Lambda is responsible for generating pre-signed POST URLs to upload files to an S3 bucket.

(cd lambdas/presign; rm -f lambda.zip; zip lambda.zip handler.py)
awslocal lambda create-function \
    --function-name presign \
    --runtime python3.11 \
    --timeout 10 \
    --zip-file fileb://lambdas/presign/lambda.zip \
    --handler handler.handler \
    --role arn:aws:iam::000000000000:role/lambda-role \
    --environment Variables="{STAGE=local}"

Create the function URL:

awslocal lambda create-function-url-config \
    --function-name presign \
    --auth-type NONE

Copy the FunctionUrl from the response, you will need it later to make the app work.

Image lister lambda

The list Lambda is very similar:

(cd lambdas/list; rm -f lambda.zip; zip lambda.zip handler.py)
awslocal lambda create-function \
    --function-name list \
    --handler handler.handler \
    --zip-file fileb://lambdas/list/lambda.zip \
    --runtime python3.11 \
    --role arn:aws:iam::000000000000:role/lambda-role \
    --environment Variables="{STAGE=local}"

Create the function URL:

awslocal lambda create-function-url-config \
    --function-name list \
    --auth-type NONE

Resizer Lambda

(
    cd lambdas/resize
    rm -rf package lambda.zip
    mkdir package
    pip install -r requirements.txt -t package --platform manylinux2014_x86_64 --only-binary=:all:
    zip lambda.zip handler.py
    cd package
    zip -r ../lambda.zip *;
)
awslocal lambda create-function \
    --function-name resize \
    --runtime python3.11 \
    --timeout 10 \
    --zip-file fileb://lambdas/resize/lambda.zip \
    --handler handler.handler \
    --dead-letter-config TargetArn=arn:aws:sns:us-east-1:000000000000:failed-resize-topic \
    --role arn:aws:iam::000000000000:role/lambda-role \
    --environment Variables="{STAGE=local}"

Connect the S3 bucket to the resizer lambda

awslocal s3api put-bucket-notification-configuration \
    --bucket localstack-thumbnails-app-images \
    --notification-configuration "{\"LambdaFunctionConfigurations\": [{\"LambdaFunctionArn\": \"$(awslocal lambda get-function --function-name resize | jq -r .Configuration.FunctionArn)\", \"Events\": [\"s3:ObjectCreated:*\"]}]}"

Create the static s3 webapp

awslocal s3 mb s3://webapp
awslocal s3 sync --delete ./website s3://webapp
awslocal s3 website s3://webapp --index-document index.html

Using the application

Once deployed, visit http://webapp.s3-website.localhost.localstack.cloud:4566

Paste the Function URL of the presign Lambda you created earlier into the form field.

awslocal lambda list-function-url-configs --function-name presign
awslocal lambda list-function-url-configs --function-name list

After uploading a file, you can download the resized file from the localstack-thumbnails-app-resized bucket.

Testing failures

If the resize Lambda fails, an SNS message is sent to a topic that an SES subscription listens to. An email is then sent with the raw failure message. In a real scenario you'd probably have another lambda sitting here, but it's just for demo purposes. Since there's no real email server involved, you can use the SES developer endpoint to list messages that were sent via SES:

curl -s http://localhost.localstack.cloud:4566/_aws/ses

An alternative is to use a service like MailHog or smtp4dev, and start LocalStack using SMTP_HOST=host.docker.internal:1025 pointing to the mock SMTP server.

Run integration tests

Once all resource are created on LocalStack, you can run the automated integration tests.

pytest tests/

GitHub Action

The demo LocalStack in CI, .github/workflows/integration-test.yml contains a GitHub Action that starts up LocalStack, deploys the infrastructure to it, and then runs the integration tests.

Contributing

We appreciate your interest in contributing to our project and are always looking for new ways to improve the developer experience. We welcome feedback, bug reports, and even feature ideas from the community. Please refer to the contributing file for more details on how to get started.

Cloud Pods

Cloud Pods are a mechanism that allows you to take a snapshot of the state in your current LocalStack instance, persist it to a storage backend, and easily share it with your team members.

You can convert your current AWS infrastructure state to a Cloud Pod using the localstack CLI. Check out our Getting Started guide and LocalStack Cloud Pods CLI reference to learn more about Cloud Pods and how to use them.

To inject a Cloud Pod you can use Cloud Pods Launchpad wich quickly injects Cloud Pods into your running LocalStack container.

Click here to launch the Cloud Pods Launchpad and inject the Cloud Pod for this application by clicking the Inject button.

Alternatively, you can inject the pod by using the localstack CLI. First, you need to download the pod you want to inject from the releases. Then run:

localstack state import /path/to/release-pod.zip