Requirement	Mechanism	Verification
Load balancer	Nginx round-robin	/whoami alternates instances
API	Spring Cloud Gateway	all traffic via :8000
Multiple instances	event-service-1/-2	both healthy, both serve
Async communication	RabbitMQ topic exchange	order.confirmed -> notification
Concurrency control	atomic DB UPDATE	2nd reserve on qty=1 -> 409
Resilience	Resilience4j CB+retry	breaker opens at failureRate=1.0
Idempotency	gateway key + consumer PK + status guard	kill-test: one charge/ticket/email
Observability	JSON logs + Prometheus + Grafana	dashboard panels + structured logs
Mocked payment/email	standalone mock + log/DB notification	admin/mode flips, /notifications

Start it

docker compose up --build

1. Prove load balancing:

curl http://localhost:8000/whoami
curl http://localhost:8000/whoami

2. Register a user:

curl -X POST http://localhost:8000/users -H "Content-Type: application/json" -d "{\"email\":\"leo@test.com\",\"name\":\"Leonardo\"}"

3. Create an event:

curl -X POST http://localhost:8000/events -H "Content-Type: application/json" -d "{\"name\":\"Show D\",\"eventDate\":\"2026-11-01T20:00:00\",\"price\":120.00,\"availableQuantity\":3}"

4. List events (note the id, usage bellow):

curl http://localhost:8000/events

5. Create an order (reserve a ticket):

curl -X POST http://localhost:8000/orders -H "Content-Type: application/json" -d "{\"userId\":1,\"eventId\":1}"

6. Pay (confirms order + publishes order.confirmed):

curl -X POST http://localhost:8000/orders/1/pay -H "Content-Type: application/json" -d "{\"paymentMethod\":\"CREDIT_CARD\",\"amount\":120.00}"

7. Check the async notification arrived:

curl http://localhost:8000/notifications

Circuit breaker demo (admin endpoint not routed through gateway, use 8090)

100% failure
curl -X POST http://localhost:8090/admin/mode -H "Content-Type: application/json" -d "{\"failureRate\":1.0}"

watch retries then breaker opening in logs
curl -X POST http://localhost:8000/orders/1/pay -H "Content-Type: application/json" -d "{\"paymentMethod\":\"CREDIT_CARD\",\"amount\":120.00}"

recover
curl -X POST http://localhost:8090/admin/mode -H "Content-Type: application/json" -d "{\"failureRate\":0.0}"

Oversell prevention demo (create event with quantity 1, reserve it twice, second must 409):

curl -X POST http://localhost:8000/events -H "Content-Type: application/json" -d "{\"name\":\"Soldout\",\"eventDate\":\"2026-12-01T20:00:00\",\"price\":50.00,\"availableQuantity\":1}"

curl -X POST http://localhost:8000/orders -H "Content-Type: application/json" -d "{\"userId\":1,\"eventId\":1}"

curl -X POST http://localhost:8000/orders -H "Content-Type: application/json" -d "{\"userId\":1,\"eventId\":1}"

http://localhost:8000 -> API Gateway (all app traffic)

http://localhost:9090/targets -> Prometheus (confirm all targets UP)

http://localhost:3000 -> Grafana (admin/admin)

http://localhost:15672 -> RabbitMQ management (tickets/tickets)

docker compose psㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤ ㅤ# status of all containers
docker compose logs -f order-serviceㅤ ㅤㅤ# follow one service's logs
docker compose logs -f payment-gateway-mock
docker compose downㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤ # stop everything
docker compose down -vㅤㅤㅤㅤㅤㅤㅤ ㅤ ㅤ # stop + wipe volumes (fresh DB)

Grafana -> Dashboards -> New -> Add visualization -> Select Prometheus -> Paste The Query -> Set a Title -> Apply

Throughput (requests per second):

rate(http_server_requests_seconds_count[1m])

Latency (95th percentile, seconds):

histogram_quantile(0.95, rate(http_server_requests_seconds_bucket[1m]))

Error rate (non-2xx responses per second):

rate(http_server_requests_seconds_count{status=~"5.."}[1m])

Custom business metric:

tickets_sold_total

scenario:

a client fires a payment, the network drops mid-flight (client never sees the response), the client retries with the same key; and the system ends with exactly one charge and one ticket.

set up fresh order curl -X POST http://localhost:8000/events -H "Content-Type: application/json" -d "{\"name\":\"Idem Test\",\"eventDate\":\"2026-12-20T20:00:00\",\"price\":90.00,\"availableQuantity\":5}" curl -X POST http://localhost:8000/orders -H "Content-Type: application/json" -d "{\"userId\":1,\"eventId\":1}"

make the gateway slow so you have time to kill the request curl -X POST http://localhost:8090/admin/mode -H "Content-Type: application/json" -d "{\"latencyMs\":8000}"

fire the payment, then hit Ctrl+C after ~2 seconds; client abandons, but the charge may have landed curl -X POST http://localhost:8000/orders/1/pay -H "Content-Type: application/json" -d "{\"paymentMethod\":\"CREDIT_CARD\",\"amount\":90.00}" ^C

Reset latency and retry the exact same payment (same order -> same idempotency key): curl -X POST http://localhost:8090/admin/mode -H "Content-Type: application/json" -d "{\"latencyMs\":100}" curl -X POST http://localhost:8000/orders/1/pay -H "Content-Type: application/json" -d "{\"paymentMethod\":\"CREDIT_CARD\",\"amount\":90.00}"

Now prove the end state; exactly one of everything: order is CONFIRMED, not double-processed curl http://localhost:8000/orders/1

exactly one email curl http://localhost:8000/notifications

inventory decremented exactly once (started at 5, should be 4) curl http://localhost:8000/events