telark
Operations

Health checks

What each service exposes for liveness and readiness.

Every telark service exposes both a liveness and a readiness probe on its standard health port. The probes are intentionally narrow so that a probe failure is always actionable.

Discovery

  • GET /healthz — process is up, Redis client is connected, and the informer cache is bootstrapped. Returns 200 once initial sync is complete.
  • GET /readyz — the leader has been elected and the discovery stream consumer group is healthy. Returns 200 only on the replica that is currently serving discovery cycles.

Exporter

  • GET /healthz — process is up, snapshot PVC is mounted, and the retention worker has registered.
  • GET /readyz — exporter can write to the PVC and reach the controller that applies rollbacks.

Auth

  • GET /healthz — process is up.
  • GET /readyz — passkey credential store and OIDC client config have loaded.

Configurator

  • GET /healthz — process is up.
  • GET /readyz — GlobalConfig CRD is reachable and the cached config is current.

Probe configuration

The bundled Helm chart wires every probe to a livenessProbe and readinessProbe with initialDelaySeconds tuned for cluster cold starts. The defaults are conservative; tune downward only after observing the steady-state pod startup time in your environment.

Environment variables

The runtime environment each telark service reads at startup.

Metrics and observability

Prometheus metrics, structured logs, and the dashboards telark emits.

On this page

DiscoveryExporterAuthConfiguratorProbe configuration