How tutora-zpi/microservices works
Project type and purpose
• This is a polyglot, domain-driven microservices system.
• It comprises several bounded contexts (“account”, “meeting”, “alerts”, plus future “ml” and “status” services) that communicate via REST, WebSocket/SSE/WebRTC, and an event bus (RabbitMQ).
• Each microservice is independently packaged (Spring Boot, NestJS, Go) and containerized with Docker Compose.
File structure
• Top-level folders:
• account (user-service, class-service)
• meeting (meeting-scheduler-service, board-service, chat-service, voice-service, voice-webrtc)
• alerts (notification-service)
• (ml / status – mentioned in README but not yet in tree)
• Each subfolder contains its own Dockerfile, env samples, and compose overrides.
• Persistence/backing stores vary by service: relational (PostgreSQL via Spring Flyway or Go migrations), NoSQL (MongoDB), Redis (for ephemeral meeting state), AWS S3 (avatars, voice blobs).
• RabbitMQ config appears in almost every service for event publishing & consuming.
• Authentication: JWT/OAuth2 in user-service, propagated by other services via RabbitMQ or via HTTP headers.
README for additional insights
• The README enumerates each domain:
– account: auth & class management
– meeting: scheduling, real-time board, chat, voice, WebRTC signaling
– alerts: notification dispatch
– ml: note generation from audio (future)