The Gin framework uses a global mode setting (gin.SetMode()) which is not
thread-safe when tests run in parallel. Removing t.Parallel() from metrics
tests that use gin.SetMode() prevents data races when running tests with
the race detector enabled.
All tests now pass with 'make test' which includes -race flag.
Story 1.2: Database Layer
- Test database client creation, connection, ping, and close
- Test connection pooling configuration
- Tests skip if database is not available (short mode)
Story 1.3: Health Monitoring and Metrics
- Test health registry registration and checking
- Test database health checker
- Test liveness and readiness checks
- Test metrics creation, middleware, and handler
- Test Prometheus metrics endpoint
Story 1.4: Error Handling and Error Bus
- Test channel-based error bus creation
- Test error publishing with context
- Test nil error handling
- Test channel full scenario
- Test graceful shutdown
- Fix Close() method to handle multiple calls safely
Story 1.5: HTTP Server and Middleware
- Test server creation with all middleware
- Test request ID middleware
- Test logging middleware
- Test panic recovery middleware
- Test CORS middleware
- Test timeout middleware
- Test health and metrics endpoints
- Test server shutdown
Story 1.6: OpenTelemetry Tracing
- Test tracer initialization (enabled/disabled)
- Test development and production modes
- Test OTLP exporter configuration
- Test graceful shutdown
- Test no-op tracer provider
All tests follow Go testing best practices:
- Table-driven tests where appropriate
- Parallel execution
- Proper mocking of interfaces
- Skip tests requiring external dependencies in short mode
- Fix error return value checks (errcheck)
- Fix unused parameters by using underscore prefix
- Add missing package comments to all packages
- Fix context key type issue in middleware (use typed contextKey)
- Replace deprecated trace.NewNoopTracerProvider with noop.NewTracerProvider
- Fix embedded field selector in database client
- Remove trailing whitespace
- Remove revive linter (as requested) to avoid stuttering warnings for public API interfaces
All linting and formatting checks now pass.
Implemented Epic 1 core kernel and infrastructure stories:
Story 1.1: Enhanced DI Container
- Added providers for database, health, metrics, and error bus
- Extended CoreModule to include all core services
Story 1.2: Database Layer with Ent ORM
- Created Ent schema for User, Role, Permission, AuditLog entities
- Implemented many-to-many relationships (User-Role, Role-Permission)
- Created database client wrapper with connection pooling
- Added database provider to DI container with migration support
Story 1.3: Health Monitoring and Metrics System
- Implemented health check registry and interface
- Added database health checker
- Created Prometheus metrics system with HTTP instrumentation
- Added health and metrics providers to DI container
Story 1.4: Error Handling and Error Bus
- Implemented channel-based error bus
- Created ErrorPublisher interface
- Added error bus provider with lifecycle management
Story 1.5: HTTP Server Foundation
- Created HTTP server with Gin framework
- Implemented comprehensive middleware stack:
- Request ID generation
- Structured logging
- Panic recovery with error bus integration
- Prometheus metrics collection
- CORS support
- Registered core routes: /healthz, /ready, /metrics
- Integrated with FX lifecycle for graceful shutdown
All components are integrated via DI container and ready for use.
