Go production best practices?

Khi đưa ứng dụng Go lên production, cần đảm bảo: structured logging (slog, zap, hoặc zerolog) dạng JSON dễ parse, graceful shutdown để không mất request đang xử lý, health check endpoints cho load balancer.

Về config, đọc từ environment variables (viper hoặc envconfig). Dockerfile dùng multi-stage build với scratch hoặc distroless để giảm attack surface và image size xuống 5-15MB. Chất lượng code: race detector (go test -race) trong CI và golangci-lint để catch lỗi sớm.

Luôn propagate context xuyên suốt request chain, cấu hình connection pooling hợp lý cho database, và expose metrics qua Prometheus để monitoring hiệu năng.

For production Go services: use structured logging (slog, zap, zerolog) outputting JSON for log aggregators; implement graceful shutdown to avoid losing in-flight requests; expose health check endpoints for load balancers.

For configuration, read from environment variables (viper, envconfig). Use multi-stage Docker builds with scratch or distroless to shrink image size to 5-15MB and reduce attack surface. Enforce quality via race detector (go test -race) in CI and golangci-lint for early detection.

Always propagate context through the full request chain, configure database connection pooling properly, and expose Prometheus metrics for performance monitoring.