Observability

Health

• GET /health (liveness)
  • Always returns 200 with {"status":"ok"} if the server is running.
• GET /ready (readiness)
  • Returns 200 with {"status":"ready"} when the DB can be reached.
  • Returns 503 with {"status":"unavailable"} when the DB is unavailable.

Example:

curl -sS http://localhost:8080/health
curl -sS http://localhost:8080/ready

Metrics

• GET /metrics (Prometheus exposition format)

By default, /metrics is public. To require Bearer auth:

export METRICS_AUTH_ENABLED=true

When auth is enabled, /metrics uses the same Bearer tokens as the API (AUTH_TOKENS / legacy AUTH_TOKEN):

TOKEN=token-current
curl -H "Authorization: Bearer ${TOKEN}" http://localhost:8080/metrics

Prometheus scraping notes:

• If you enable auth, prefer supplying the token via bearer_token_file (or similar) rather than embedding it directly in config.
• Alternatively, leave METRICS_AUTH_ENABLED=false and restrict /metrics at the network/reverse-proxy layer.

Request IDs

mjr.wtf propagates X-Request-ID:

• If the request includes X-Request-ID, it is reused.
• Otherwise a UUID is generated.

The value is echoed back on responses via the X-Request-ID header and included in request logs as request_id.

Logging

Logging is structured via zerolog.

• LOG_LEVEL (default: info)
• LOG_FORMAT (json or pretty, default: json)
• LOG_STACK_TRACES (default: true; include stack traces in panic recovery logs)

Each HTTP request is logged with fields like request_id, method, path, status, size, and duration.