Skip to content

Nodes

One card per node, each with three meters and a row of condition pills.

Meter What it shows
CPU Allocatable vs used, as a percentage and absolute cores.
Memory Allocatable vs used.
Pods Scheduled pods against the node’s pod capacity.

The pod meter is the one people forget. A node with spare CPU and memory that has hit its pod cap cannot take another pod, and the resulting Unschedulable reads as a capacity problem when it is a limit.

Kubernetes node conditions — Ready, MemoryPressure, DiskPressure, PIDPressure, NetworkUnavailable — each shown as a pill, plus node role chips (control-plane, worker) and the node’s age.

Node version skew is one of the six upgrade gates. Nodes more than one minor behind the control plane after a hop are a blocker, and this tab is where you see the spread before you plan it.

Terminal window
rtz get no
rtz get no -o wide