Back to all scenarios
Scenario #324
Storage
Kubernetes v1.22, GCP Filestore
Pod CrashLoop Due to Read-Only Volume Remount
Pod volume was remounted as read-only after a transient network disconnect, breaking app write logic.
Find this helpful?
What Happened
During a brief NFS outage, volume was remounted in read-only mode by the NFS client. Application kept crashing due to inability to write logs.
Diagnosis Steps
- 1Checked mount logs: showed NFS remounted as read-only.
- 2kubectl describe pod: showed volume still mounted.
- 3Pod logs: permission denied on write.
Root Cause
NFS client behavior defaults to remount as read-only after timeout.
Fix/Workaround
• Restarted pod to trigger clean remount.
• Tuned NFS mount options (soft, timeo, retry).Lessons Learned
NFS remount behavior can silently switch access mode.
How to Avoid
- 1Monitor for dmesg or NFS client remounts.
- 2Add alerts for unexpected read-only volume transitions.