Back to all scenarios
Scenario #304
Storage
Kubernetes v1.21, Azure Disk CSI
PVC Not Rescheduled After Node Deletion
A StatefulSet pod failed to reschedule after its node was deleted, due to Azure disk still being attached.
Find this helpful?
What Happened
A pod using Azure Disk was on a node that was manually deleted. Azure did not automatically detach the disk, so rescheduling failed.
Diagnosis Steps
- 1Pod stuck in ContainerCreating.
- 2CSI logs showed "Volume is still attached to another node".
- 3Azure Portal confirmed volume was attached.
Root Cause
Manual node deletion bypassed volume detachment logic.
Fix/Workaround
• Detached the disk from the Azure console.
• Recreated pod successfully on another node.Lessons Learned
Manual infrastructure changes can break Kubernetes assumptions.
How to Avoid
- 1Use automation/scripts for safe node draining and deletion.
- 2Monitor CSI detachment status on node removal.