Twitter API

Twitter is simple Go API which sends the messages and query for messages containing a specific tag.

Choices

1. Fiber - REST API framework, chosen beacuse of good performance benchmarks
2. MongoDB - document database, perfect for this case
3. Python/pytest - for integration/E2E tests
4. Kubernetes - perfect to deploy and scale application
5. Skaffold/Tilt - nice k8s development tools

Endpoints

POST /messages

Example: http://localhost:3000/messages

JSON Schema:

1. text - message to be posted
2. tags - tags to be assigned to message

Sample JSON payload:

{
    "text": "sample message",
    "tags": ["tag1", "tag2"]
}

Sample response:

{
    "count": 1,
    "result": {
        "id": "60c9f412a8b77a729096d3a5",
        "text": "sample message",
        "tags": [
            "tag1",
            "tag2"
        ],
        "created_at": "2021-06-16T12:52:34.278183643Z"
    },
    "status": true
}

GET /messages

Example: http://localhost:3000/messages?from=2020-06-13T15:00:05Z&to=2021-06-13T15:00:05Z&tags=zx,zs

Query parameters:

1. tags - list of tags, comma separated
2. from - RFC3339-compatible start filtering time
3. to - RFC3339-compatible stop filtering time

Sample response:

{
    "count": 1862,
    "status": true
}

Development

1. Install Skaffold
2. Install kubectl
3. Install minikube

All needed dependencies and quickstart here.

4. skaffold dev
5. Use uploaded Postman collection (Twitter.postman_collection.json) to call API

Linting and testing

Prerequisites: use configuration from Configuration section.

1. Unit tests Type following commands to execute unit tests in Twitter development container:

   • make dev
   • make test

2. Linting Folowing command installs necessary test dependencies and launches linters(golangci-lint, golint, goimports, gofmt) and checkers (gocyclo, go vet):

   • make dev
   • make lint

3. E2E tests To launch E2E tests enter test Docker container (assumption: Twitter app, database and Test container are in the same network):

   • make dev - Enter development container
   • make build; ./twitter_app - in dev container
   • make dev-tests - to enter test container
   • pytest - to launch tests

4. Coverage report In order to generate coverage report, type following commands:

   • make dev - enter development container
   • make coverage - generate global coverage report
   • make coverhtml - generate HTML coverage report

Configuration

Both Twitter API and Test container are configured using environment variables(it's convenient to have .env file in main application directory):

API_PORT=3000
DB_URI=mongodb://root:root@localhost:27017/?authSource=admin&readPreference=primary&appname=MongoDB%20Compass&ssl=false
DB_NAME=twitter
DB_COLLECTION_NAME=messages

Deployment

Kubernetes is used to deploy solution:

1. Build prod Docker image: make build_image
2. Apply deployment yaml: kubectl apply -f deployment/twitter-deployment.yaml
3. Check your minikube IP: minikube ip
4. You can access Twitter API on minikube IP address, port 32000, like http://192.168.49.2:32000

Summary

Shortcuts:

1. No authentication
2. No performance tests (Python and locust could be used to measure that)
3. No security tests
4. No monitoring solution (ELK/Prometheus)
5. No CI (private repository issues)
6. No api/v1 used in endpoints - all in all, it's JSON API ;-)
7. No helm and other k8s improvements
8. No automated API docs like Swagger