Skip to content

Resource context

scan prints a compact cluster resource summary alongside the diagnostic output, giving you the context needed to interpret findings without switching to a separate tool.

What is included

  • Allocatable — total CPU and memory the cluster can schedule onto.
  • Reserved / Requests — the sum of all pod resource requests.
  • Limits — the sum of all pod resource limits.
  • Live usage — actual node-level consumption, when metrics-server is installed.

Live usage is best-effort: without metrics-server the summary still shows allocatable, reserved, and limits, and notes usage as unavailable.

OOMKill annotations

Each OOMKilled finding is annotated with the killed container's requests and limits. This gives you the data you need to judge whether to raise the limit or scale out, right in the scan output, without a separate kubectl describe command.

Interaction with --explain

The resource summary is included in what is sent to --explain so the model can give memory- and CPU-aware advice — for example, suggesting a limit increase versus a horizontal scale-out based on cluster headroom.

Example output

Cluster resources
  CPU    allocatable: 24 cores   requests: 14.2 cores   limits: 28.0 cores
  Memory allocatable: 96 Gi      requests: 38.4 Gi      limits: 64.0 Gi      usage: 41.2 Gi

Note

Reading live usage requires a single GET on the metrics API and is always read-only. Without metrics-server the usage column shows (unavailable).