Cluster
The Cluster page shows the Kubernetes resources underneath HEAT. Kubernetes is the container orchestration system that runs every HEAT service and processing workload. Open Monitoring, then Cluster (/admin/monitoring/cluster).
Three cards summarise the cluster: Nodes ready, Pods running against the total, and Persistent volumes with their combined size.
Pods
A pod is one containerised workload. The page lists every pod in the cluster, including system services, so you can see everything that is running. Filter boxes narrow the list by namespace, pod name and node.
| Column | Meaning |
|---|---|
| Namespace | The Kubernetes namespace. HEAT runs in the heat namespace |
| Pod | The pod name. The prefix names the service or runner |
| Status | Running, Pending, or a failure state |
| Node | The cluster node the pod runs on |
| CPU, MEM | Live usage where the cluster exposes it |
| READY | Ready containers against the total for the pod |
| RST | Restart count |
| AGE | How long the pod has existed |
Two signals matter most:
- A pod not in Running, or a READY count below its total.
- A climbing RST count, which is the first sign of a crash loop.
Every pod in the heat namespace is part of HEAT. The image behind a pod tells you which version of a service is deployed. Confirm the core services, the Core API, the v2 API, HEAT Auth and the Control Centre, are running and on the expected image tag.
Nodes and volumes
The summary cards report node readiness and the persistent volume claims behind HEAT. A persistent volume claim is the durable disk a service asked Kubernetes for. Services report their own usage, so the volumes view shows allocation. The Monitoring overview blob store gauge covers how full the object store is.
Pod logs
Click a pod to open its live log viewer at /admin/monitoring/cluster/pods/.... The viewer streams the pod’s output. Pause freezes the stream, and a save control downloads the visible log to a file for sharing.
Reading a pod’s logs is the quickest way to see why a service is unhealthy after the Cluster table flags it.