Observability

conduvia-observability — Health, metrics, distributed tracing, structured logs, and dashboards. Observability is core, not optional.

Core — Always Deployed

conduvia-observability

Every component emits structured data. Correlating that data across the full delivery path is a first-class concern. Ingress metrics cover request rate, validation failure rate, schema rejection rate. Worker metrics cover delivery success/failure, retry distribution, DLQ routing rate. The Conduvia console provides ops/admin-facing views for all operational data.

Full architecture reference →

Prometheus + Grafana metrics pipeline
W3C Trace Context (traceparent) end-to-end propagation
Structured JSON logging with correlation-id and tenant
Per-alias, per-tenant, per-connector metric dimensions
AI-facing alias labels: token consumption, MCP scope distribution
Trace correlation search across full delivery path
Circuit breaker state dashboards per delivery target