Orders Service Backend (Redis + RabbitMQ)

Flask + SQLAlchemy API with Redis caching and asynchronous writes via RabbitMQ. Reads stay fast with cache; writes are queued and applied by a background worker to avoid DB spikes while keeping read-your-own-writes consistency.

Prerequisites

• Python 3.11+
• Redis (e.g. docker run -d --name redis -p 6379:6379 redis:7)
• RabbitMQ / Amazon MQ (AMQP) reachable with the credentials in .env (local docker-compose provided)
• MySQL/RDS (optional; defaults to SQLite)

Setup

python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env  # fill in DB, Redis, RabbitMQ, admin secrets

Key .env variables:

• DATABASE_URL or DB_USER/DB_PASS/DB_HOST/DB_NAME
• REDIS_URL (default redis://localhost:6379/0)
• ORDERS_CACHE_TTL (default 30), TEMP_ORDER_CACHE_TTL (default 60)
• ADMIN_PASSWORD
• RABBITMQ_HOST, RABBITMQ_PORT, RABBITMQ_USER, RABBITMQ_PASSWORD, RABBITMQ_QUEUE_NAME, MQ_USE_SSL

Using AWS ElastiCache (Redis) + Amazon MQ

• Point REDIS_URL to your ElastiCache endpoint, e.g. REDIS_URL=rediss://<primary-endpoint>:6379/0 (use rediss for TLS).
• Amazon MQ (AMQP):
  • RABBITMQ_HOST=<amazon-mq-broker-endpoint>
  • RABBITMQ_PORT=5671 for TLS/AMQPS (use 5672 if non-TLS)
  • RABBITMQ_USER / RABBITMQ_PASSWORD = your MQ user credentials
  • MQ_USE_SSL=True when using TLS (recommended)
  • RABBITMQ_QUEUE_NAME=order_write_jobs (keep default unless you renamed it)
• When pointing to cloud Redis/MQ, do not start the local docker compose up redis rabbitmq.

Running (recommended)

From back_end/:

# ensure Redis and RabbitMQ are up first
docker compose up -d redis rabbitmq  # from repo root, optional helper
./start_services.sh

• API: gunicorn --workers 3 --threads 4 --bind 0.0.0.0:8080 (override with API_WORKERS, API_THREADS)
• Worker count default: WORKER_COUNT=4
• PID/log files live next to the scripts (server.pid, worker_*.pid, server.log, worker_*.log)
• Access logs: server-access.log captures each request (override via ACCESS_LOG_FILE, ACCESS_LOG_FORMAT)
• Startup connectivity logs show Redis/RabbitMQ/DB reachability in server.log (and worker logs) when the app boots

Stop everything:

./stop_services.sh

Manual run (if you prefer):

# start Redis and RabbitMQ separately (e.g., docker or cloud broker)
python run.py                  # API (dev server)
python worker.py               # queue consumer

Write + Read Flow (async + cache)

• POST /orders: validate -> write full payload to Redis temp_order:{id} (60s TTL) -> enqueue job to RabbitMQ -> return 202 + order_id. No direct DB write.
• Worker (worker.py): consumes queue; performs SQL INSERT/UPDATE/DELETE with simulated 0.3s lag; invalidates Redis list/detail caches and lets the temp entry expire naturally to preserve the read-your-own-writes window.
• GET /orders/<id>: check temp_order:{id} first (read-your-own-writes); else cached detail orders:detail:{id}; else DB.
• PATCH /orders/<id>/status and DELETE /orders/<id>: enqueue jobs; return 202.
• GET /orders: cache key orders:list:{version} with version bumped on writes.

Admin Endpoints

• POST /admin/reset — wipes all orders/items (requires ADMIN_PASSWORD)
• POST /admin/seed — seeds fake orders ({ "count": 100 } + admin password)

Troubleshooting

• Verify brokers: redis-cli PING, rabbitmqctl list_queues (or AWS MQ console)
• If cache is disabled: set CACHE_DISABLED=true
• Adjust temp cache TTL: TEMP_ORDER_CACHE_TTL

Load Testing

Locust file: tests/locustfile.py

locust -f tests/locustfile.py --host=http://localhost:8080