Datastore Free Space Not Updated After Disk Deletion
Article ID: 421990
Updated On:
Products
Issue/Introduction
VMDK disks have been deleted and the datastore free space has not updated to reflect this change. The disks appear as removed in the Datastore, ESXi, and on the NetApp volume.
A refresh and rescan at the VM, ESXi, datastore, and volume levels, but the free space remains unchanged.
Environment
ESXi 7.x and above
VCenter 7.x and above
Cause
vCenter/ESXi Perspective (Logical View):
• VMware generally reports space from a logical perspective.
• When you create a thin-provisioned VMDK, vCenter deducts the full provisioned size from the available space in its metadata, even if the VM has only written a small amount of data.
• For example, if you create a 300 GB thin-provisioned disk, vCenter views 300 GB as "potentially used" space that is no longer freely available to other VMs on that datastore, to prevent over-allocation issues from the VMware side.
• Once a VM writes data to a block, that block is considered "in use" by the hypervisor and remains so, even if the data is deleted within the guest OS.
• Other files, such as VM swap files (which are created when a VM is powered on and are sized according to the VM's assigned memory), also contribute to the space vCenter marks as used.
NetApp Perspective (Physical View):
• The NetApp storage array, using its own thin provisioning capabilities (often deployed in FlexVol volumes), reports the actual physical disk space consumed by data written to the LUN.
• The array is only concerned with the blocks that have physically received data (actual 1s and 0s). If a 300 GB VMDK only has 80 GB of actual data written to it, the NetApp array will correctly show only about 80 GB as used physical space.
• Space-saving technologies like deduplication and compression on the NetApp side can further reduce the physical space used, making the physical usage even lower than what ESXi sees.
Resolution
Check NetApp Snapshots: If the snapshots exist, delete the snapshot holding the VMDK.
When you delete a VMDK from vCenter on an NFS datastore, vSphere sends a standard file delete command (unlink) to the NetApp array. The file is removed from the active file system immediately.
If you have NetApp Snapshots enabled on that volume, the "deleted" VMDK is still being referenced by the most recent snapshot. The blocks cannot be freed until that snapshot expires or is deleted.
vSphere shows the file is gone. NetApp shows the volume is still full (or the usage moved from "Active Data" to "Snapshot Reserve").
Browse Datastore: Confirm the .vmdk file is actually gone and wasn't just removed from the inventory list. ( From what I can see it has been removed )
Check permissions: Ensure the Root Squash settings on the NetApp Export Policy allow the ESXi host to perform delete operations (usually requires Superuser: sys or Any).
Feedback
