Trident CSI NodeUnstage Failure: iSCSI Session Logout Occurs Before LUN Cleanup
Applies to
- NetApp Trident CSI Driver (versions prior to 25.02.1, especially 23.07 and 24.02)
- Kubernetes clusters using iSCSI-backed Persistent Volumes
- Ubuntu 22.04 worker nodes (multipath enabled)
- Multi-attach scenarios with iSCSI
- Environments where Trident manages iSCSI session lifecycle
Issue
During Persistent Volume (PV) detach (NodeUnstage) operations, Trident logs out an iSCSI session shared by multiple volumes before all dependent LUNs are fully cleaned up. This causes subsequent NodeUnstage operations for other volumes using the same session to fail with device discovery errors, leaving PVs stuck and requiring host reboots for recovery.
Symptoms:
- Kubelet logs show persistent UnmountDevice failures and “multi-attach” errors.
- Trident logs repeatedly show:
level=warning msg="Could not get device info for removal, skipping host removal steps." error="scannot completed for LUN6 on target iqn.1992-08.com.netapp:sn.c07af9cb5e4111f084a9d039eaad9561:vs.71"
- Trident debug logs:
Logging out of session [sid:61,target:iqn...,portal:10.90.224.11,3260]Logout of [sid:61,target:iqn...,portal:10.90.224.11,3260] successful....Could not get device info for removal, skipping host removal steps. error="scannot completed for LUN6 on target iqn..."
- PVs remain in “multi-attach” state and cannot be reclaimed.
- Host reboots are often required to clear stuck devices.
