Microservices

A basic example of microservice architecture which demonstrates communication between a few loosely coupled services and deployment to AWS.

• Written in Go
• Uses RabbitMQ to communicate between services
• Uses WebSocket to talk to the front end
• Stores data in PostgreSQL
• Stores cache in Redis
• Uses React for front end development
• Builds with Docker
• Deployed on AWS with CloudFormation templates
  • ECS using EC2
  • AWS Fargate
• Uses CI/CD pipeline

Local use

To run the example, clone the Github repository and start the services using Docker Compose. Once Docker finishes downloading and building images, the front end is accessible by visiting localhost:8080.

git clone https://github.com/ebosas/microservices
cd microservices

docker-compose up

Database

To inspect the database, launch a new container that will connect to our Postgres database. Then enter the password demopsw (see the .env file).

docker run -it --rm \
    --network microservices_network \
    postgres:13-alpine \
    psql -h postgres -U postgres -d microservices

Select everything from the messages table:

select * from messages;

Redis

To inspect Redis, connect to its container via redis-cli.

docker run -it --rm \
    --network microservices_network \
    redis:6-alpine \
    redis-cli -h redis

Get all cached messages or show the number of messages.

lrange messages 0 -1
get total

RabbitMQ

Access the RabbitMQ management interface by visiting localhost:15672 with guest as both username and password.

Back end

To access the back end service, attach to its docker container from a separate terminal window. Messages from the front end will show up here. Also, standart input will be sent to the front end for two way communication.

docker attach microservices_backend

Deployment to Amazon ECS/AWS Fargate

cd deployments and create the CI/CD pipeline stack. Once finished, visit the ExternalUrl available in the load balancer's Outputs tab in CloudFormation.

aws cloudformation create-stack \
	--stack-name MicroservicesFargate \
	--template-body file://pipeline.yml \
	--parameters \
		ParameterKey=DeploymentType,ParameterValue=fargate \
		ParameterKey=EnvironmentName,ParameterValue=microservices-fargate \
		ParameterKey=GitHubRepo,ParameterValue=<github_repo_name> \
		ParameterKey=GitHubBranch,ParameterValue=<github_branch> \
		ParameterKey=GitHubToken,ParameterValue=<github_token> \
		ParameterKey=GitHubUser,ParameterValue=<github_user> \
	--capabilities CAPABILITY_NAMED_IAM

Github repo setup

Fork this repo to have a copy in your Github account.

Then, on the Github access token page, generate a new token with the following access:

• repo
• admin:repo_hook

Deleting stacks

When deleting the ECS cluster stack (cluster-ecs.yml) in CloudFormation, the auto scaling group needs to be manually deleted. You can do it from the Auto Scaling Groups section in the AWS EC2 console.

With capacity providers, container instances need to be protected from scale-in. This interferes with the automatic deletion process in CloudFormation.

References

Deployment is based on these templates: https://github.com/nathanpeck/ecs-cloudformation

Local development

For development, run the RabbitMQ and Postgres containers with Docker Compose.

docker-compose -f docker-compose-dev.yml up

Generate static web assets for the server service by going to web/react and web/bootstrap and running:

npm run build-server

React

For React development, run npm run serve in web/react and change the script tag in the server's template to the following:

<script src="http://127.0.0.1:8000/index.js"></script>