OK Observability — Evidence-Driven Cluster Visibility
Metrics. Dashboards. Logs. Alerts. Verified per cluster.
OK Observability is the OpenKubes Observability Capability, defined by ADR-Platform-018. Rather than treating observability as an optional collection of dashboards, OpenKubes defines a testable contract for local cluster visibility.
Explore OK Observability
| Topic | What you will learn |
|---|---|
| Observability Stack | Reference implementation: Prometheus, Grafana, Alertmanager and OpenSearch |
| Capability Contract | The five mandatory guarantees and their verification boundaries |
| Installation & Verification | How ok-cluster consumes the capability and runs the gated install |
Contract over components
The ok-observability repository owns the contract, profiles, charts, dashboards, alert rules and tests. ok-cluster consumes the capability during cluster operations. Implementation profiles may vary by environment without silently changing the contract.
The standard implementation is designed to work per cluster without cross-cluster connectivity, including environments with restricted or air-gapped networks when their prerequisites are satisfied.
The capability's maintained repository README reports successful contract-test runs on two independent clusters (ok-shared and ok-robotics). That evidence demonstrates those deployments, not universal readiness of every possible cluster or envelope.
Architecture and ownership
OpenKubes platform decisions (ADR-Platform-018 / 024 / 027)
|
v
ok-observability: capability contract + implementation profiles + tests
|
v
ok-cluster: explicit gated install + cluster-specific provider values
|
v
Target cluster: metrics + dashboards + logs + alerting
|
v
Contract result + bounded evidence
Start with the Observability Capability repository and its repository guide. The site is an entry point; the source repositories remain authoritative.