Announcement
This page is also available in Deutsch, Español, Français, Italiano, 日本語, and Português.
Unevictable Pods view in PerfectScale InfraFit
See exactly which pods are keeping your nodes alive and blocking scale-down. The new Unevictable Pods view in PerfectScale InfraFit highlights these workloads across any node group, with per-node resource breakdowns for CPU, memory, and GPU. Pair it with node recommendations to improve bin-packing, get more out of Karpenter or Cluster Autoscaler, and lower your Kubernetes spend.
By andrii
Unevictable Pods view in PerfectScale InfraFit
PerfectScale's Node Group view now includes an Unevictable Pods view. It surfaces the workloads blocking nodes from being removed from your cluster.
Unevictable pods are a common scale-down blocker. They reduce bin-packing efficiency and inflate cloud costs, even when autoscalers like Karpenter or Cluster Autoscaler are in place. This view makes them easy to find and act on.
What you can do
- Spot scale-down blockers fast. The top panel summarizes total running pods and unevictable pods across the selected node group.
- Drill into each node. Instance charts show resource bars for CPU, RAM, and GPU, plus a pod treemap. Unevictable pods are outlined in red.
- Switch resource views. Toggle between CPU, Memory, and GPU to analyze pressure by resource type.
- Review historical snapshots. Use the date picker to load any snapshot from the past 30 days.
- Inspect details. Click a node or pod to open a sidebar with requests, limits, actual usage, labels, and annotations.
How to access
Open InfraFit, select a node group, and switch to the Unevictable Pods tab. The latest snapshot loads by default.
Use this view alongside node recommendations to improve bin-packing, increase autoscaler effectiveness, and reduce Kubernetes spend.