Virtual Machines Report Persistent “File Busy” or “Locked” Errors After NFS Datastore Interruption in Unsupported Nested vSAN File Services Configurations

Products

• VMware vSAN
• VMware vSphere ESXi

Issue/Introduction

• VM power-on fails with: Module VMMon power on failed. Could not power on virtual machine: Timeout.
• Specific error in vmkernel.log indicating persistent open failures: NFS41: NFS41FileOpOpenFile:####: Open of obj 0x############ name -flat.vmdk (fhID 0x########) failed: Busy
• Continuous Ioctl[file] failed or LookupAndOpen[file] failed messages in hostd.log despite no active processes using the files.
• High latency warnings often precede the failure: WARNING: NFS: ####: NFS volume average I/O latency ####(us) has exceeded threshold 10000(us).

Environment

NFS datastore backed by vSAN File Service

Cause

Events disruption the NFS connectivity like APD/PDL create lock deletions often fail to resolve these 'Busy' states because the underlying storage provider (vSAN File Service) is itself a consumer of the same cluster resources. This circular dependency prevents the storage layer from correctly identifying that a lock is orphaned, as the 'locking' entity and the 'provider' entity share the same host-level identity.

 

Resolution

vSAN File Services is intended for file-sharing protocols (CIFS/NFS) for guest OS access, not for hosting VMDKs on datastores mounted back to ESXi. 

Refer documentation for best practices and configuration for vSAN File Services