Feature/observability

[Howard-nolan]

Change Summary:

• Observability: Integrated Zap for structured, level-based JSON logging to improve debugging capabilities.
• Monitoring: Added a Prometheus middleware to track key "Golden Signals" (request latency and traffic volume) and exposed them via a new /metrics endpoint.
• Local Dev Stack: Created a docker-compose.yml file to spin up a local monitoring stack (Prometheus and Grafana) alongside the application.
• Performance: Implemented a caching layer to reduce database load for frequently accessed URL redirects.

How It Works:

• Metrics Middleware: A new MetricsMiddleware wraps the HTTP router. It captures the start time of every request and records:

• http_requests_total (Counter): Labeled by method, route, and status.

• http_request_duration_seconds (Histogram): Captures latency buckets for P99/P95 calculations.

• Route Normalization: The middleware uses parameterized route names (e.g., /v1/r/{code}) rather than raw paths to prevent high-cardinality issues in Prometheus.

• Docker Stack:

• Prometheus: Configured to scrape host.docker.internal:8080/metrics every 5 seconds.

• Grafana: Pre-provisioned to visualize the Prometheus data on port 3000.

• Logging: Replaced standard fmt print statements with a global Zap logger configuration.

Additional Notes (optional):

• Access: Grafana is available at http://localhost:3000 (default creds: admin/admin) when running docker compose up.

• Trade-offs: Deferred OpenTelemetry (tracing) implementation as the current architecture is single-service and basic structured logging/metrics provide sufficient visibility for now.

• Verification: Verified that the /metrics endpoint is correctly outputting Prometheus-formatted text with appropriate buckets.