Back to all scenarios
Scenario #333
Storage
Kubernetes v1.22, OpenEBS Jiva CSI

CSI Driver Crash Loops on VolumeAttach

CSI node plugin entered CrashLoopBackOff due to panic during volume attach, halting all storage provisioning.

Find this helpful?
What Happened

VolumeAttachment object triggered a plugin bug—CSI crashed during RPC call, making storage class unusable.

Diagnosis Steps
  • 1Checked CSI node logs — Go panic in attach handler.
  • 2Pods using Jiva SC failed with AttachVolume.Attach failed error.
  • 3CSI pod restarted every few seconds.
Root Cause

Volume metadata had an unexpected field due to version mismatch.

Fix/Workaround
• Rolled back CSI driver to stable version.
• Purged corrupted volume metadata.
Lessons Learned

CSI versioning must be tightly managed.

How to Avoid
  • 1Use upgrade staging before deploying new CSI versions.
  • 2Enable CSI health monitoring via liveness probes.
Previous ScenarioNext Scenario