Chat Auth Service

This is a simple chat authentication service that use OAuth2 to authenticate users. The service is written in Go and uses Postgres for store user data and Redis for session management.

Getting Started

1. Create a .env file in the root of the project with the following variables:

   PORT=8080
   APP_ENV=local
   DB_HOST=localhost
   DB_PORT=5444
   DB_DATABASE=auth
   DB_USERNAME=<DB_USERNAME>
   DB_PASSWORD=<DB_PASSWORD>
   DB_SCHEMA=public
   SONAR_TOKEN=<SONAR_TOKEN> # Optional
   GOOGLE_APPLICATION_KEY=<GOOGLE_APPLICATION_KEY>
   GOOGLE_APPLICATION_SECRET=<GOOGLE_APPLICATION_SECRET>
   GITHUB_APPLICATION_KEY=<GITHUB_APPLICATION_KEY>
   GITHUB_APPLICATION_SECRET=<GITHUB_APPLICATION_SECRET>
   SESSION_SECRET=<SESSION_SECRET> # Any random string with 32 characters
   SESSION_MANAGER_SECRET=<SESSION_MANAGER> # Any random string with 32 characters
   REDIS_HOST=<REDIS_HOST> i use redislabs.com
   REDIS_PORT=<REDIS_PORT>
   REDIS_PASSWORD=<REDIS_PASSWORD>
   SESSION_TIMEOUT=<SESSION_TIMEOUT> # in seconds '3600s'
   

2. Run the following command to start the Postgres and Redis containers

   make docker-run

3. Run the following command to start the application

   make run

4. Using a browser navigate to localhost:8080/login/google?username=<USERNAME> to authenticate with Google.

   Obs.: github provider is not working.

5. After authenticating, will return a token that can be used to authenticate with the chat service, that is not implemented yet.

   Obs.: The token is not a JWT token, it is a random key that is stored in Redis. The value of the key is the user information, like roles and permissions.

6. Refresh and Logout endpoints still are in development.

Decision logs

• 2024/07/*: Session manager storage must be a key-value database with a ttl mechanism. First option: redis
• 2024/07/*: I decided deploy the server on aws environment. I will use dynamodb as storage to minimize the distance between the server and session manager repo.
• 2024/07/31: dynamodb is not so similar with redis as i initially think. Dynamodb ttl mechanism is not automatic and the storage is not in-memory. A better redis equivalent inside aws it would be ElasticCache, but elasticCache is not on localstack free tier, which difficult integration test. Because of that, i will use Redis. Redis docker on local env and Redislab on cloud env.
• 2024/08/04: deploy on aws lambda is complicated and demands specific implementation. I choose deploy on aws ecs, where the implementation is not so coupled

MakeFile

run all make commands with clean tests

make all build

build the application

make build

run the application

make run

Create DB container

make docker-run

Shutdown DB container

make docker-down

live reload the application

make watch

run the test suite

make test

clean up binary from the last build

make clean