Health And Readiness

CloudGrid services expose liveness and readiness probes. Readiness returns unhealthy while required dependencies are unavailable or a service is draining.

Probe Table

Service	Default port	Liveness	Readiness
BFF	3000	/livez	/readyz, /api/health
OTLP collector HTTP	4318	/livez	/readyz
OTLP collector gRPC	4317	Reported by HTTP /readyz	Reported by HTTP /readyz
storage-read	8081	/livez	/readyz
storage-write	8082	/livez	/readyz
control-plane	8084	/livez	/readyz
AI eval runner	8085	/livez	/readyz

Local Checks

curl -fsS http://localhost:3000/readyz
curl -fsS http://localhost:4318/readyz
curl -fsS http://localhost:8081/readyz
curl -fsS http://localhost:8082/readyz
curl -fsS http://localhost:8084/readyz

Readiness Dependencies

diagram

Common Readiness Failures

Symptom	Likely cause	Check
BFF ready but GraphQL times out	Private service not subscribed to NATS subject	storage-read and control-plane logs
storage-read not ready	SurrealDB unavailable or schema readiness failed	SurrealDB logs and storage-read /readyz
collector rejects ingest	invalid auth, content type, size, or bridge publish failure	collector logs and NATS readiness
control-plane not ready	SurrealDB schema or configured self-observability project validation failed	control-plane logs

Health Versus Readiness

Liveness answers whether the process should keep running. Readiness answers whether the process should receive traffic.

Do not treat /livez as a dependency check. Use /readyz for service routing and local troubleshooting.

Next Step

Inspect private bridge behavior with Message bridge operations.