Pending and memory failures

Opening your lesson

Start a resource incident with its status and events. Pending means the Pod has not reached a runnable placement; requests, node selection or storage may block it. OOMKilled means a running container exceeded its memory allowance. Restarting either workload without changing the cause repeats the failure. This local cluster has one small deterministic node and fictional memory-demand annotations so both paths can be reproduced safely. Use describe to identify the boundary, then apply a request or limit that fits the stated workload. The tasks keep CPU reservation and runtime memory allowance distinct, with a second wrong value that looks close but still can't succeed.

printf '%s' 'apiVersion: "v1"
kind: "Pod"
metadata:
  name: "cobalt-197"
spec:
  containers:
    - name: "cobalt-197"
      image: "nginx:1.27"
      ports:
        - containerPort: 80
      resources:
        requests:
          cpu: "500m"
          memory: "256Mi"
        limits:
          memory: "512Mi"
' > 'workload.yaml'
kubectl delete pod cobalt-197
kubectl apply -f workload.yaml

A Pod stays Pending when no node can satisfy its requests. Adjusting only the limit doesn't change the scheduler request.

Choose your character

Your Work Gym account