Dashboard

The Dashboard is a single-page overview of the whole cluster - health, workload status, capacity, usage trends and anything that needs attention - so you can answer “is my cluster OK right now?” at a glance before drilling into any one resource.

Open it from Features ▸ Dashboard at the top of the sidebar. It fills the window (no detail panel of its own); clicking anything on it navigates you to that resource’s normal view, and the back / forward buttons include the Dashboard so you can step back to it.

Health banner

The banner at the top summarises the cluster as Healthy, Needs attention or Critical, with a one-line explanation. It is derived from what the Dashboard itself flags below - so it always agrees with the Active Problems list.

• Healthy - nothing is flagged.
• Needs attention - softer issues only: a node under memory/disk/PID pressure, or a Deployment that isn’t fully rolled out.
• Critical - a pod is crashing/failed, or a node is NotReady.

Pods that are merely Pending (a Deployment or CronJob just spinning up) are treated as transient and do not flip the banner - a job coming up doesn’t make the cluster unhealthy.

Status

Three tiles summarise the most important counts. Each tile is clickable and jumps to the matching list.

• Pods - total, broken down into Running, Pending, Issues and Succeeded.
• Deployments - total, with Available vs Degraded, plus StatefulSet / DaemonSet counts.
• Nodes - total, with Ready, NotReady, Cordoned and Pressure counts.

Cluster usage

Live CPU and memory trends for the whole cluster, aggregated across all nodes. The dashed line is allocatable capacity, the green line is total requested, and the blue line is actual used. The charts need a few metric samples to appear, so they fill in over the first minute after connecting. (Charts honour your CPU / memory unit preferences - see Display Units.)

Capacity

How much of the cluster is committed and consumed right now:

• Requested - the sum of pod resources.requests, as a percentage of allocatable. This is what the scheduler has reserved.
• Used - actual consumption reported by metrics-server.

Bars turn amber above 75% and red above 90%.

Reclaimable

If you’ve run Right-Sizing analyses, this tile estimates how much CPU and memory you could give back by adopting the recommended requests - summed across every analysed Deployment, multiplied by its replica count. It’s a quick measure of cluster-wide over-provisioning. Click it to open Right-Sizing. (Empty until you’ve run at least one analysis - see Right-Sizing.)

Top consumers

The heaviest pods by CPU and by memory in the current scope, so you can spot what’s eating the cluster. Click any row to open that pod.

Active problems

A live list of everything the Dashboard considers a problem - crashing or failed pods, NotReady or pressured nodes, and degraded Deployments. Each row opens the resource. When everything is healthy it shows a single “nothing needs attention” line.

Namespace scope

The Dashboard respects the namespace selector in the toolbar. When you scope to a single namespace, the pod, workload, top-consumer and problem figures reflect that namespace - but node capacity and the usage graphs stay cluster-wide, because nodes belong to the whole cluster. A note under the title reminds you when a namespace filter is active. (With several namespaces selected, the Dashboard shows cluster-wide figures.)