DevOpsPadDownload

Check Pod CPU and memory usage

Open 监控 (Monitoring) under 可观测 (Observe) in the Infra sidebar. It shows a snapshot of current CPU and memory usage for each Pod of the selected application, read from the cluster’s existing metrics-server: CPU in mCPU (1,000 mCPU = 1 core), memory in MiB, and the time the sample was taken.

It is a point-in-time view, not a monitoring system. DevOpsPad stores no metrics history, and metrics are never written to the project directory.

The monitoring page (监控) showing total CPU in mCPU, memory in MiB and ready Pods, with a per-Pod table and sample times

What you see

  • Summary tiles: CPU 使用量 (CPU usage), 内存使用量 (memory usage) and Pod 就绪 (Pods ready).
  • A table with Pod, CPU（mCPU）, 内存（MiB） (memory) and 采样时间 (sample time) for each Pod.

Tick 每 10 秒刷新 (Refresh every 10 seconds) in the toolbar to poll, or refresh manually.

Requirements

  • metrics-server must be running in the cluster. K3s bundles it by default; a cluster started with it disabled shows 暂无可用指标 (no metrics available) and a note that metrics-server and read permission are required.
  • Your identity needs list on pods in the metrics.k8s.io/v1beta1 API for the environment’s namespace. See RBAC permissions.

How Pods are matched

DevOpsPad matches Pods using the application’s full label selector from project.yaml, including matchExpressions. When the API supports it, it first requests a lightweight Table index and then fetches only the candidates; otherwise it falls back to a full list.

Within the same application page, results are reused: switching back within 15 seconds shows the cached result, and after 15 seconds the last result stays visible while DevOpsPad refreshes in the background. The footer shows when the data was read. A manual refresh, a config change or a release invalidates the cache. Nothing is cached across app restarts.

Limits

  • No historical time series, graphs or alerts.
  • Node-level usage is not shown.
  • Lists are capped at 500 resources.

Frequently asked questions

Why does the Monitoring page say no metrics are available?

The cluster needs metrics-server installed, and your identity needs list permission on pods in metrics.k8s.io/v1beta1. K3s ships metrics-server by default unless it was disabled.