Kubernetes
Kubernetes in Owlpane is the cluster-aware layer of infrastructure monitoring. The cluster hub and explorers live under /observability/kubernetes, with a Kubernetes 360 entry in the Infrastructure navigation at /observability/infrastructure/kubernetes. From a cluster you can drill into nodes, pods, workloads, services, ingresses, and certificates.
What it is
A Kubernetes cluster changes faster than any static host inventory: pods are scheduled and evicted, workloads roll out, services re-point, and certificates expire. Monitoring it well means seeing metrics, events, logs, and incidents against the objects they belong to, not just as a flat stream of series.
Owlpane links its Kubernetes projects to clusters, so the same cluster name that appears in stored metrics resolves to a project you can open, explore, and scope across signals.
Key capabilities
- Clusters list — every cluster linked to a Kubernetes project, plus any unlinked cluster telemetry that still appears in stored metrics.
- Cluster health — unhealthy pods, container restarts, CPU and memory as a percentage of allocatable, and Kubernetes events (warning and normal) for the whole cluster or a namespace.
- Explorer — a cluster explorer for browsing the objects a cluster contains.
- Object detail — per-object pages for nodes, pods, workloads, services, ingresses, and certificates.
- Unified signals — cluster logs, metrics, traces, and incidents reachable from the hub, so an alert can be followed into the object that caused it.
Where it lives
| Surface | Console route | What it does |
|---|---|---|
| Kubernetes 360 | /observability/infrastructure/kubernetes | Cluster hub for cluster logs, metrics, traces, and incidents. |
| Kubernetes | /observability/kubernetes | Cluster list and cluster health. |
| Explorer | /observability/kubernetes/explorer | Browse the objects in a cluster. |
| Node detail | /observability/kubernetes/nodes/:node | Per-node metrics and state. |
| Pod detail | /observability/kubernetes/pods/:namespace/:pod | Per-pod metrics, restarts, and events. |
| Workload detail | /observability/kubernetes/workloads/:kind/:namespace/:name | Per-workload rollouts and health. |
| Service detail | /observability/kubernetes/services/:namespace/:name | Per-service endpoints and traffic. |
| Ingress detail | /observability/kubernetes/ingresses/:namespace/:name | Per-ingress routing state. |
| Certificate detail | /observability/kubernetes/certificates/:namespace/:name | Per-certificate validity state. |
Plans
The number of Kubernetes clusters is available on all plans. Restricting a member to specific Kubernetes namespaces is available on Scale and Enterprise. See Plan and billing.
How data gets in
Deploy the Owlpane Kubernetes agent with Helm — a cluster collector plus a per-node DaemonSet — to collect pod logs, kubelet and container metrics, Kubernetes events, and cluster OTLP traces, then forward them to your workspace. Telemetry RBAC is read-only by default. Cluster names link to Kubernetes projects so the surfaces above can resolve them.
- Kubernetes Helm agent — install, values, verification, and troubleshooting.
- Kubernetes integration packs — dashboard packs for cluster telemetry.
- Metric ingestion — the kubelet and container series behind the charts.
- Agent samples vs device fleets — the cluster agent's sample catalog versus IoT fleets.
Concepts
| Term | Meaning |
|---|---|
| Cluster | A linked group of nodes reporting kubelet and container metrics. |
| Namespace | A logical partition inside a cluster. |
| Workload | A Deployment, StatefulSet, or similar controller managing pods. |
| Node | A machine that runs pods. |
| Pod | The smallest schedulable unit, one or more containers. |
| Ingress / Certificate | Routing rules and TLS validity for services exposed outside the cluster. |
How the surfaces relate
The cluster list is the entry point: it resolves cluster names to Kubernetes projects, including any stored telemetry that has no project yet. From a cluster, the hub links across signals — cluster logs, metrics, traces, and incidents — and the object pages go one level deeper, so an alert can be followed into the node, pod, or workload that caused it.
Related
- Infrastructure — the fleet view the cluster hub sits inside.
- Traces — spans from workloads instrumented by the agent.
- Logs — container and application logs collected cluster-wide.
- Alerts and on-call — cluster rules and incident routing.
- Carbon — energy and emissions attributed to cluster nodes and namespaces.
Last updated