Pod has unbound immediate PersistentVolumeClaims. preemption: 0/1 nodes are available: 1 No preemption victims found for incoming pod

General Discussions
1

I’ve been experimenting with the hack/local_up_cluster.sh and making progress. I am deploying perhaps more non-trival development helm charts now. I am currently deploying a chart and seeing Warning FailedScheduling 20s default-scheduler 0/1 nodes are available : pod has unbound immediate PersistentVolumeClaims. preemption: 0/1 nodes are available: 1 No preemption victims found for incoming pod.. for a pending pod. I thought perhaps I should create a local PV and PVC and it might resolve this. So I created that and then killed the pod but it didn’t resolve the issue upon restart. Might someone suggest what next steps I can take in my learning process?

1 Like
2

Hi, [Colin_Williams]

Please check the steps below step by step

  1. Check Node Availability: Ensure that you have enough available nodes in your cluster. If not, consider adding more nodes.
  2. Resource Limits: Verify that nodes have sufficient resources (CPU, memory) to meet the PVC requirements. Adjust resource limits if needed.
  3. Storage Classes: Confirm that the StorageClasses in your PVCs are available and compatible with your cluster’s storage backend.
  4. Pod Priority and Preemption: If using Pod priority and preemption, ensure correct priority levels and preemption policies.
  5. Node Affinity and Anti-affinity: Check for node affinity or anti-affinity rules that may prevent pod scheduling.
  6. Pod Disruption Budgets: If PodDisruptionBudgets are set, ensure they allow for the eviction of pods necessary for preemption.
  7. Check Logs: Examine scheduler, controller-manager, and kubelet logs for error messages. Review events associated with PVCs and Pods for more details.
    …
    Remember to consider your cluster’s specific details and configurations while troubleshooting. If you need more guidance or have specific details about your cluster, feel free to share it .
3

Thanks. I no longer am testing locally this as it was posted a year ago. But perhaps your suggestion will assist someone else.