vSphere Replication fails with "Prune disks" error and 'Invalid argument' due to NFS APD events
Article ID: 444857
Updated On:
Products
Issue/Introduction
- Enhanced Replication mapping tests remain stuck in a "Testing" state.
- Protected virtual machines show a status of Not Active.
- Replication tasks fail with disk pruning errors and file locks.
In the vSphere Replication UI or hms.log, you may see errors similar to:
A replication error has occurred on the vSphere Replication server for replication 'VM_NAME'. Details: 'Error for (datastoreUUID: "UUID"), (diskId: "RDID-UUID"), (pathname: "VM_NAME/disk.vmdk")ObjLib error: Invalid argument; Failed to combine files ... Prune disks could not remove disk instance (instanceKey=####); While completing partial prune'.
Environment
vSphere Replication 9.x
Cause
This issue is caused by intermittent NFS storage disconnects (All Paths Down / APD events) on the underlying storage layer.
When the storage path fails, the Host-Based Replication (HBR) stream over port 32032 is interrupted. These disconnects prevent the ESXi hosts from performing critical I/O operations on the VMDK files, such as disk consolidation and pruning. This results in stale file locks and a hang in the replication health check (mapping test) as the services cannot reach a consistent state on the impacted datastores.
Resolution
To restore replication functionality, the underlying storage must be stabilized before refreshing the replication services.
- Work with your storage vendor to investigate and resolve the root cause of the NFS APD events. Ensure that all ESXi hosts have stable, consistent paths to the replication datastores.
- Once storage stability is confirmed, restart the replication management service to clear the stale "Testing" states.
#systemctl restart hms
Additional Information
Disk prune errors may also have other causes, see: vSphere Replication for virtual machines are in error state with the consolidation error: "Prune disks could not remove disk instance"
Feedback
