Instance overview

Shared infrastructure and status for a FireboltInstance.

A FireboltInstance provides the shared infrastructure its Engines need:

• PostgreSQL stores Metadata Service data. Use the built-in single-node database for evaluation, or configure an external PostgreSQL database for production.
• The Metadata Service provides account and Engine metadata.
• The Envoy Gateway routes client queries to ready Engine pods.

Engines in the namespace wait until their referenced Instance publishes the required Metadata Service endpoint and Instance ID.

For the complete API surface, see the FireboltInstance CRD reference.

Instance lifecycle

Phase	Meaning
Provisioning	One or more components are being created or are not ready.
Ready	The Metadata Service and Gateway are ready.
Degraded	A component that was ready is currently unavailable.
Failed	The Instance needs manual intervention. Inspect its conditions and events.

Authentication and TLS have dedicated conditions: AuthReady, EngineTLSReady, and GatewayTLSReady. Inspect these conditions even when the Metadata Service and Gateway are available.

Inspect Instance status

kubectl get fire -n firebolt

NAME         PHASE   GATEWAY   METADATA   AGE
production   Ready   true      true       24h

For conditions, endpoints, and events:

kubectl describe fire production -n firebolt
kubectl get fire production -n firebolt -o yaml

Customize component pods

Use spec.gateway.template and spec.metadata.template to configure user-owned pod settings such as scheduling, labels, annotations, image pull secrets, resources, and sidecars. The Firebolt Operator owns component identity, ports, probes, commands, required volumes, and other fields needed for correct operation.

See Firebolt Operator-owned component fields before adding template fields.

Select the Engine metrics transport

The Firebolt Operator reads Engine query metrics for graceful drain checks and auto-stop:

• spec.metricScrapeMode: PodIP is the default and connects directly to each Engine pod on its metrics port.
• spec.metricScrapeMode: ApiserverProxy routes requests through the Kubernetes API server. Set Helm value rbac.apiserverProxyGrant=true when you use this mode.

Use ApiserverProxy only when the Firebolt Operator cannot reach pod IPs directly and your cluster permits API-server proxy traffic to the Engine metrics port.

Worked manifests are available in the examples/ directory.