Operations

Back to index

Health Endpoints

The control plane exposes:

• GET /healthz
• GET /readyz
• GET /readyz/{subsystem}
• GET /api/v1/system/healthz
• GET /api/v1/system/readyz
• GET /api/v1/system/readyz/{subsystem}

Liveness

/healthz is the simple process-level check.

Readiness

/readyz returns a JSON report covering subsystem checks such as:

• database
• cache
• platform settings
• profiles
• telemetry
• agents
• launchable suites

Required subsystems can make readiness fail with 503 Service Unavailable.

Middleware And Request Discipline

The shared HTTP stack adds:

• CORS handling
• request IDs
• session context population
• tracing hooks
• HTTP metrics
• audit middleware

This keeps cross-cutting behavior consistent across the API surface.

Telemetry

OpenTelemetry can be enabled with the OTLP environment variables in .env.

Common settings:

• OTEL_EXPORTER_OTLP_ENDPOINT
• OTEL_SERVICE_NAME
• OTEL_EXPORTER_OTLP_INSECURE
• OTEL_EXPORTER_OTLP_HEADERS
• OTEL_RESOURCE_ATTRIBUTES

If telemetry is not configured, the readiness report marks that subsystem as disabled instead of hard failing the server.

Cache Layer

Redis is optional. When configured, it is used for:

• cached reads
• favorites and workspace acceleration
• execution runtime cache
• platform and profile cache

If Redis is missing, the control plane can still run, but the readiness report will reflect the cache state.

Datastores

The primary datastore can be:

• MongoDB
• PostgreSQL

MongoDB is the default local path in the checked-in .env.

Environment Inventory

The environments page is backed by a runtime inventory service that:

• polls Docker resources
• tracks orchestrator process liveness
• identifies zombie environments
• supports SSE updates
• allows per-environment or global cleanup

Worker Health

The worker process exposes its own GET /healthz endpoint and a small runtime API for:

• info
• run
• cancel
• cleanup

That lets a control plane or external operator verify worker availability independently.