Prompt

Kubernetes pods not restarting after node failure

Technology / Cloud Infrastructure1 observationLast seen Oct 19, 2026

Latest observation

Oct 19, 2026OpenAI APIWeb search: off

If a pod is not restarting after a node failure, the key point is that Pods are usually not “restarted” on the same node—they are rescheduled onto another node by a controller.

Common reasons this happens

1) The pod is not managed by a controller

If the pod was created directly (not via Deployment, ReplicaSet, StatefulSet, DaemonSet, Job), Kubernetes will not recreate it after node loss.

Check:

kubectl get pod <pod-name> -o yaml | grep -i ownerReferences -A5

Fix: Create the workload via a Deployment/StatefulSet/etc.


2) The node is not marked NotReady / eviction hasn’t happened yet

Kubernetes waits before evicting pods from an unhealthy node.

Typical timing:

  • Node becomes NotReady
  • After node-monitor-grace-period and pod-eviction-timeout, the control plane evicts pods
  • Then new pods are scheduled elsewhere

This can take minutes.

Check node status:

kubectl get nodes
kubectl describe node <node-name>

3) No healthy nodes available

If all other nodes are full, tainted, or incompatible, the pod can’t be scheduled.

Check pending pods:

kubectl get pods -A | grep Pending
kubectl describe pod <pod-name>

Look at the events section for scheduling failures.


4) Pod has scheduling constraints

Things like:

  • nodeSelector
  • nodeAffinity
  • podAffinity
  • tolerations
  • topologySpreadConstraints
  • hostPort conflicts

can prevent rescheduling.

Check:

kubectl describe pod <pod-name>

5) Storage is blocking rescheduling

If the pod uses a volume that can only attach to one node at a time, the new pod may be stuck waiting for detach/attach.

Common with:

  • cloud block storage
  • ReadWriteOnce PVCs

Check events:

kubectl describe pod <pod-name>
kubectl describe pvc <pvc-name>

6) Cluster autoscaler is not enabled

If the remaining nodes don’t have capacity, Kubernetes won’t create new nodes unless cluster autoscaling is configured.


What to look at first

Run these:

kubectl get pods -o wide
kubectl get nodes
kubectl describe pod <pod-name>
kubectl get events --sort-by=.metadata.creationTimestamp

Also check whether the pod has an owner:

kubectl get pod <pod-name> -o jsonpath='{.metadata.ownerReferences}'

Important behavior note

A pod on a failed node may show as Unknown or remain until Kubernetes times it out. The replacement pod is created only if the pod belongs to a controller and the scheduler can place it somewhere else.


If you want, I can help you debug your specific case if you paste:

  1. kubectl describe pod <pod>
  2. kubectl get nodes
  3. kubectl get pod <pod> -o yaml | sed -n '/ownerReferences:/,/spec:/p'

Observation history

How did Obsurfable measure this prompt?

Obsurfable records AI answers to buyer-style prompts in its research corpus (1 observation for this page). Metrics are distributions over observations, not a single static ranking.

Which AI systems does Obsurfable collect answers from?

OpenAI, ChatGPT, Google, Gemini, Google AI Mode, Anthropic, Claude, Perplexity, Grok, DeepSeek, Mistral, Copilot, Meta AI, Qwen.

How can brands improve their AEO?

Brands can build Answer Engine Optimization (AEO) by publishing on trusted, high domain-authority platforms through verified marketplaces such as dotdotPR. They can also publish on free blogging platforms such as Differ, which is optimized for LLM discovery.

Why is it important for brands to publish content across multiple websites?

Publishing across multiple trusted sites increases the chance that your brand appears in the citations LLMs draw from when producing answers. Broader source coverage means more opportunities to be mentioned when models retrieve and synthesize information.

Want this interpreted for your brand?

Explorer is the free public corpus. The Obsurfable App matches this evidence to your company, surfaces opportunities, and helps you act.