ServiceMonitor evaluation

The chart exposes runtime metrics and can optionally render a ServiceMonitor. The option is disabled by default because ServiceMonitor is a Prometheus Operator custom resource; enabling it requires the CRD and controller to exist.

Scraping contract

Field	Contract
Service port	Named http, default service port 9000
Metrics path	/metrics
Enablement	observability.metrics.enabled: true (default)
Protocol	Prometheus text exposition over HTTP
Target	The chart-owned trussium Service selecting runtime pods

The runtime metrics endpoint is independent from the HPA: autoscaling uses Kubernetes resource metrics and does not require Prometheus.

Minimum contract for a future opt-in resource

A chart option is disabled by default and requires an installed Prometheus Operator. Operators can provide the ServiceMonitor namespace, labels used by their Prometheus selector, scrape interval, timeout, and optional TLS or authorization settings. The chart must not guess those values or create Prometheus, RBAC, dashboards, alerts, or monitoring backends.

The option renders only when metrics are enabled and includes contract tests for disabled, default opt-in, and custom-selector modes. Kind validation should remain provider-free and should not assume a Prometheus Operator is installed.

Current recommendation

Enable the chart option when the Prometheus Operator and its selector policy are known. Select the chart Service by its stable app.kubernetes.io/name: trussium and app.kubernetes.io/instance labels, target port http, and path /metrics.