tzurot-observability

Structured logging with correlation IDs enables end-to-end request tracing across services. Implement Pino for JSON-formatted logs, attach a requestId to each request, and propagate it through all downstream calls. This allows you to filter and correlate logs by request, pinpointing failures and bottlenecks in production.

Cross-service debugging requires correlation IDs to link logs, metrics, and errors to a single request flow. Expose health and metrics endpoints to monitor service dependencies in real time. Use structured logging to capture rich context—userId, requestId, error objects—making it faster to diagnose where and why failures occur.

Apply privacy rules when structuring logs to exclude or redact sensitive fields like passwords, tokens, and PII. Use the golden logging pattern with explicit field selection rather than string interpolation, which makes it harder to accidentally log secrets. Document which fields are safe to log in your codebase.

Yes, health and metrics endpoints work with Railway to expose real-time service status and dependency checks. Implement `/health` endpoints to report readiness and dependency health, and `/metrics` endpoints to track performance. Railway logs and CLI commands can then query these endpoints to diagnose and monitor service behavior.

Logging captures detailed events and errors with full context for debugging specific incidents. Metrics aggregate counts, latencies, and health signals for trends and alerting. Together they provide both the granular detail to troubleshoot production issues and the aggregate view to detect systemic problems early.

JSON-formatted structured logs work best with Railway's log parsing and filtering. Pino outputs JSON by default, making logs queryable and filterable in Railway's CLI and dashboard. Avoid string interpolation; use explicit structured fields so Railway tools can reliably extract, search, and correlate logs.