Updated August 24, 2026
The cluster page shows the following without any setup or port-forwarding:
-
Overview / KPI strip — nodes ready, pods, CPU and memory across the cluster, plus control-plane health: etcd latency p99 and 24-hour restart counts. Green means the control plane is healthy; sustained etcd latency above 100 ms or restarts indicate a problem on our side that we are alerted on.
-
Workloads — everything running in the cluster, read live from the cluster’s API server. Read-only: you deploy and change workloads with kubectl, Helm or GitOps; this tab is for tracking. Five sub-tabs:
- Deployments / StatefulSets / DaemonSets — each row shows the namespace, Ready count (
2/3 = 2 of 3 replicas ready), the container image with its tag, age, and a status: Healthy (all replicas ready), Pending (not ready yet, but created less than 10 minutes ago — still starting), Degraded (not all replicas ready past that window — worth investigating: check the pod’s logs in the Logs tab).
- Services — type, cluster IP and ports of every Service.
- Ingresses — the HTTP routes into the cluster and their hosts.
Filter by name or namespace to find a specific workload. A workload stuck Degraded with a rising restart count in the Logs tab is the usual sign of a crash loop or a failing image pull.
- Metrics — live utilization snapshot from the cluster’s metrics-server: cluster-wide CPU, memory, pods and readiness, with a per-node breakdown. Refreshed every 60 seconds.
- Pods & logs — live pod list with streaming logs (tail, since, previous container).
- Nodes — per-node readiness (Ready / Not ready / Cordoned), CPU and memory.
- Activity — every action on the cluster from anyone in your org (dashboard or API), plus lifecycle events: provisioning steps, scaling, autoscaler actions, warnings. Retained for 90 days, exportable as CSV.
The same data is available on the metrics and events API endpoints. For monitoring inside the cluster, install Grafana or kube-prometheus-stack from App Hub. Last modified on August 24, 2026