vCenter Datastore Browser continues to show files which have been deleted on an NFS datastore

Products

• VMware vSphere ESXi
• VMware vCenter Server

Issue/Introduction

Users may observe a discrepancy where the vSphere Client (vCenter) datastore browser continues to list files, folders, or Virtual Volumes (VVols) that have been previously deleted or unmounted. This mismatch can interfere with storage reporting and the creation of new replication scenarios.

Upon reviewing the vvold.log files on the affected ESXi hosts, several successful unmount and unbind operations were confirmed. The logs may show:

• "VVolDevice:PrepareDeviceForBatachUnbind Preparing device for unbind"
• "Successfully acquired user lock on VVolDevice"
• "VVolDevice::UnmountDevice successfully unmounted"
• "VVolDevice:: FinishBatchUnbind Finishing unbind for rfc####.############"

Despite these successful status messages from the host perspective, the objects remained visible in the vCenter inventory, indicating that the metadata or physical deletion had not been finalized at the storage layer. The capacity still shows as used as well.

 

Environment

• VMware vCenter Server
• VMware ESXi
• Storage Type: VVol-backed datastores or External Storage Arrays (e.g., HPE)

Cause

This issue occurs when the ESXi host successfully sends and processes unmount or unbind commands, but the underlying storage array fails to physically remove the objects or update the metadata on the destination datastore.

While the host logs (vvold.log) show successful completion of tasks such as VVolDevice::UnmountDevice successfully unmounted, the storage layer retains the stale entries. vCenter continues to display these objects because it has not received a final confirmation of deletion from the storage provider's management layer.

Resolution

To resolve this discrepancy, follow these steps:

1. Validate Host Logs: Review the vvold.log on the affected ESXi hosts to confirm that unbind and unmount tasks were successful. Look for "Successfully acquired user lock" and "FinishBatchUnbind" messages.
2. Verify Filesystem: Connect to the ESXi host via SSH and run the ls -lh command on the affected datastore path (/vmfs/volumes/[Datastore_Name]) to confirm the files are physically gone from the host perspective.
3. Storage Rescan: Perform a manual storage rescan in vCenter on all hosts connected to the datastore to refresh the inventory metadata.
4. Engage Storage Vendor: If the files still appear in vCenter after a rescan, contact your storage hardware vendor (e.g., HP Support) to perform a cleanup of the stale objects directly on the storage array.
5. Re-sync Replication: Once the storage vendor has cleared the stale data, you may proceed with creating new replication scenarios or storage tasks.