Observability

If you can't debug a problem in production from the dashboards, the dashboards aren't done.

The three signals

• Logs — structured JSON, request ID + trace ID propagated. Used for forensic debugging.
• Metrics — RED (Rate, Errors, Duration) per endpoint and per critical operation. Used for trends and alerting.
• Traces — for any request that crosses a service boundary. Used for latency debugging.

How a service loads its log and metric settings (config file, plus DEPLOY_ENV for the deployment label): Logs and metrics configuration.