Skip to main content
Pods lists every pod in the namespace you picked, failing ones first. Open it from the sidebar, with G then P, or with ⌘3 in the desktop app (Ctrl3 on Windows and Linux).
The Pods list, with failing pods at the top showing CrashLoopBackOff and ImagePullBackOff, and their restarts.The Pods list, with failing pods at the top showing CrashLoopBackOff and ImagePullBackOff, and their restarts.

Every pod, its status and restarts, with the failing ones first.

​
The list

The status is the most specific reason KubeStacks can find: a container’s waiting reason (like ImagePullBackOff), why a pod failed (like OOMKilled or Evicted), or why it can’t be scheduled. Hover it for the message behind it. Health and status lists them all.

​
Open a pod

Click a row, or move to it with ↑ ↓ and press ↵.
A pod's details next to the list: its containers, their usage and state.A pod's details next to the list: its containers, their usage and state.

A pod's details next to the list: its containers, their usage and state.

Overview starts with the facts: the node it runs on, its pod and host IPs, QoS class, service account, restart policy, and Controlled by, a link to the ReplicaSet, Job or other object that manages it. Then a card for each container, init containers first:
  • Its image and ports.
  • Its state: Running for 2h, Running for 5m, not ready, the reason it’s waiting (hover for the message), or how it ended, like Error (exit code 1).
  • How many times it restarted.
  • Its CPU and memory use right now, next to its requests and limits.
  • Last terminated, when it has restarted: the reason and exit code of its previous run. This is how an OOMKilled restart shows on a pod that’s running again.
After the containers come the pod’s conditions, labels and annotations.

​
The other tabs

​
What you can do

From the panel, the ⋯ menu (.), the row’s right-click menu or the command palette:
  • Shell and Debug…, to look inside. A debug container works even in images with no shell.
  • Forward a port…, to reach the pod from your computer. Desktop app
  • Restart, for a pod something manages: it’s deleted and its controller makes a new one.
  • Evict…, through the Eviction API, so PodDisruptionBudgets are respected.
  • Delete…, with Force for a pod stuck terminating.
Each one is described in Actions by kind. Right-click a pod for Copy kubectl logs too, alongside Copy name and Copy kubectl describe.

​
Thousands of pods

Lists load in chunks of 500 and only draw the rows on screen, so a cluster with thousands of pods scrolls as smoothly as one with ten. Long lists refresh less often, and a list stops at 5,000 objects with a note saying so. Use a label selector to see the rest, or raise the limit with KUBESTACKS_MAX_LIST_ITEMS.

Logs

A pod’s logs, or every pod of a workload, merged.

Health and status

What each pod status means, and how it’s worked out.