VMFS Datastore Inaccessible with Error Missing LVM Magic and No Signature Found

Products

• VMware vSphere ESXi

Issue/Introduction

A VMFS datastore on a Fibre Channel LUN becomes inaccessible and fails to mount on VMware ESXi.

The following errors appear in /var/log/vmkernel.log:

 

LVM: 6504: Could not open device naa.################################:1 ... Permission denied
Vol3: 2629: Could not open device 'naa.###############################:1' for volume open: Not found

Executing the vSphere On-disk Metadata Analyzer (voma) command returns:

ERROR: Missing LVM Magic. Disk doesn't have valid LVM Device
No signature for LVM nor VMFS was found.

Symptoms:

• Running partedUtil getptbl shows that the GPT partition table is intact, but the VMFS filesystem headers within the partition are corrupted or unreadable.

• The datastore cannot be mounted or recognized by the ESXi host.

Environment

ESXi 8.x 

Cause

The Logical Volume Manager (LVM) magic header and VMFS superblock signatures on the storage device partition were overwritten or corrupted, preventing the ESXi VMkernel from recognizing the logical volume structure.

Resolution

 

• If lost data must be recovered and no recent backups exist, do not write any new data or create new filesystems on the LUN. Engage Broadcom-approved Data Recovery Services prior to taking further action. Refer to Data Recovery Services for Data Not Recoverable by VMware Support for engagement details.

• If data recovery is not required or data has already been safely backed up, clear the corrupted partition table using partedUtil:

  partedUtil delete /vmfs/devices/disks/<device_naa_id> <partition_number>
  

• In the vSphere Client, navigate to Storage > New Datastore and re-format the LUN as a new VMFS 6 datastore.

• Restore virtual machine workloads from your verified backup repository.