Monitor Health and Metrics
Use /health for liveness and /ready for traffic and rollout gates.
Endpoint Map
Section titled “Endpoint Map”| Service | Port | Health | Readiness | Metrics |
|---|---|---|---|---|
| API | 3000 | /health | /ready | /metrics |
| STS | 8080 | /health | /ready | /metrics, /metrics.json |
| Gateway | 8081 | /health | /ready | /metrics, /metrics.json |
| Audit | 9090 | /health | /ready | /metrics, /metrics.json |
| Coordinator | 4000 | /health | /ready | /metrics |
Published metrics require Authorization: Bearer <METRICS_BEARER>. Keep them private even when authenticated.
Procedure
Section titled “Procedure”- Scrape every enabled service with the metrics secret.
- Track readiness, DB pool, policy age/compile errors, provider refresh, audit lag/DLQ/tamper/replay, outbox, STS circuit, and revocation freshness.
- Correlate logs and audit by request ID.
- Gate deployment on readiness plus canary exchange, Gateway call, and audit lookup.
For Compose run caracal status --ready; from a checkout use bash infra/scripts/smokeTest.sh. For Helm, verify Jobs, workloads, ServiceMonitor discovery, and rules.
Recovery
Section titled “Recovery”If health passes but readiness fails, preserve the readiness reason and inspect its dependency/safety latch. Restore the cause and drain backlogs. Preserve evidence before replacing an audit_evidence_lost process.
Next Step
Section titled “Next Step”Route signals with Configure Alerts.

