Back to all scenarios
Scenario #373
Storage
Kubernetes v1.24, default EmptyDir
Pod Reboots Erase Data on EmptyDir Volume
Pod restarts caused in-memory volume to be wiped, resulting in lost logs.
Find this helpful?
What Happened
Logging container used EmptyDir with memory medium. Node rebooted, and logs were lost.
Diagnosis Steps
- 1Post-reboot, EmptyDir was reinitialized.
- 2Logs had disappeared from the container volume.
Root Cause
EmptyDir with medium: Memory is ephemeral and tied to node lifecycle.
Fix/Workaround
• Switched to hostPath for logs or persisted to object storage.Lessons Learned
Understand EmptyDir behavior before using for critical data.
How to Avoid
- 1Use PVs or centralized logging for durability.
- 2Avoid medium: Memory unless necessary.