Dead Space Reclamation Issue on Pure Storage LUNs After VSS Backups

Products

• VMware vSphere ESXi

Issue/Introduction

This article explains why "dead space" or phantom utilization increases on storage arrays (such as Pure Storage) when running VSS-based backups (like Cohesity) on thick-provisioned VMDKs. It provides steps to reclaim space and prevent future accumulation.

Environment

 

• VMware vSphere ESXi 7.x / 8.x
• Thick Provisioned VMDKs (Eager or Lazy Zeroed)
• Cohesity or other VSS-integrated backup solutions
• Pure Storage or other VAAI-compliant arrays

 

Cause

Guest-initiated SCSI UNMAP commands (0x42) are not passed through the ESXi hypervisor to the storage array for virtual disks in a Thick Provisioned format. When VSS shadow copies are deleted within the Guest OS, the blocks are logically freed but remain marked as "allocated" on the physical storage LUN.

Resolution

To reclaim space and resolve the passthrough limitation:

1. Manual Reclamation: Run the following command to reclaim blocks currently free at the VMFS layer:  #esxcli storage vmfs unmap --volume-label=#### --reclaim-unit=200
2. Convert to Thin Provisioning: Perform a Storage vMotion of the affected Virtual Machines and select Thin Provision as the virtual disk format. Thin-provisioned disks allow Guest-OS UNMAP commands to pass through to the array.
3. Enable Guest UNMAP: After conversion, ensure the VM advanced setting disk.scsiUnmapAllowed is set to True.

Additional Information

 

• For instructions on retrieving logs, see How to collect logs for VMware vSphere.
• To speak with a customer representative or a Support Engineer see Contact Support. Scroll to the bottom of the page and click on your respective region.