VMware ESXi and vSphere Cluster Management

Delete or Unmount a VMFS Datastore in VMware ESXi

Learn when to delete or unmount a VMFS datastore in ESXi, the safety checks required, and how to avoid losing virtual machine data.

VMware ESXi provides two different ways to remove a VMFS datastore from use: delete and unmount. These actions have very different consequences. Choose based on whether the data on the underlying storage must be preserved.

VMFS Datastore Removal: Delete or Unmount?

VMFS is the VMware Virtual Machine File System used by ESXi datastores. A datastore is a storage container presented to ESXi for virtual machine files, virtual disks, ISO files, templates, and other content.

Deleting a datastore is destructive. Unmounting is a host-level disconnection. Deleting removes the VMFS datastore structure and makes its contents unavailable to every ESXi host that can access the datastore. Unmounting removes access only from the selected ESXi host while leaving the data on the storage device.

Operation	Effect on stored data	Effect on ESXi host access	Effect on other hosts	Typical use case	Key prerequisites
Delete	Removes the VMFS filesystem metadata; datastore contents become unavailable.	Removes the datastore from the host.	Affects every host that accesses the datastore.	Permanently retiring storage after all required content has been migrated or backed up.	Confirm the correct datastore, shut down affected VMs, and back up required content.
Unmount	Preserves the VMFS data on the storage device.	Disconnects the datastore from the selected ESXi host.	Other hosts can retain access if their configuration and storage connectivity permit it.	Removing one host's access or preparing storage for maintenance without destroying data.	No VMs or management dependencies may remain on the datastore; remove listed dependencies.

Delete a VMFS Datastore

Deleting is the destructive datastore-removal method. The operation removes the VMFS datastore structure and causes the files on that datastore to disappear from hosts that can access it. Do not treat deletion as a simple inventory cleanup.

Before deletion, back up or migrate every item that must be retained, including:

Virtual machines and their virtual disks
ISO files and installation media
Templates
Snapshots or other virtual machine files
Scripts, configuration files, and other non-VM content

Deletion is appropriate only when the datastore and its contents are no longer needed, or when required content has been safely moved and the storage is being permanently decommissioned. If another ESXi host still needs the datastore, deletion is not the correct action.

Delete workflow

Review the ESXi or vSphere storage inventory and identify the intended VMFS datastore by name and storage details.
Identify all virtual machines whose files or virtual disks are located there.
Shut down affected virtual machines before removing the datastore, then migrate or back up their files.
Check for ISO files, templates, scripts, and operational dependencies that also require preservation.
Confirm whether the datastore is shared by multiple ESXi hosts.
Select the datastore removal action and choose delete only after confirming that permanent removal is intended.
Review all confirmation prompts carefully before completing the operation.

Unmount a VMFS Datastore

Unmounting removes an ESXi host's access to a datastore without erasing the VMFS data on the underlying storage device. This is suitable when the contents must remain available for other hosts or for later use.

Common reasons to unmount include removing one host's access to shared storage, preparing storage for maintenance, or detaching a datastore without intentionally destroying its contents.

Unmounting is not allowed merely because a virtual machine is powered off. No virtual machines may remain on the datastore. Move the virtual machines and their files to another datastore, or otherwise remove the datastore dependency, before attempting the unmount.

Unmount workflow

Select the intended VMFS datastore in the ESXi or vSphere host storage inventory.
Verify that the selected host is the host from which access should be removed.
Move or remove all virtual machines from the datastore and shut down affected workloads during the change.
Remove the datastore from any datastore cluster.
Disable Storage I/O Control for the datastore.
Ensure the datastore is not being used for vSphere HA heartbeat storage.
Ensure Storage DRS is not using the datastore.
Choose unmount, review the dependency warnings, and confirm the operation.

Unmount Readiness Checklist

Check	Required state	Why it matters
Virtual machines	No virtual machines remain on the datastore.	VM files and virtual disks create an active dependency. A powered-off VM still resides on the datastore.
Datastore cluster	The datastore is removed from any datastore cluster.	A datastore cluster manages its member datastores together and can prevent unmounting.
Storage I/O Control	Storage I/O Control is disabled.	Storage I/O Control manages storage latency and allocation among virtual machines and must not remain active for this operation.
vSphere HA heartbeat	The datastore is not selected for HA heartbeat storage.	vSphere HA uses heartbeat files during host isolation or network partition scenarios.
Storage DRS	Storage DRS is not using the datastore.	Storage DRS manages placement and balancing of virtual machine files across datastores.
Other content	Required ISO files, templates, scripts, and other files are migrated or backed up.	Unmounting preserves data, but the selected host will not be able to access it while unmounted.

Important Terms

VMFS: VMware Virtual Machine File System, the datastore filesystem used by ESXi hosts.
Datastore: A storage container presented to ESXi for virtual machine files and other host-accessible content.
Delete datastore: A destructive operation that removes the VMFS datastore structure and makes its files unavailable to hosts that can access it.
Unmount datastore: An operation that removes one ESXi host's access without erasing the data stored on the underlying device.
Datastore cluster: A group of datastores managed together, commonly with Storage DRS.
Storage I/O Control: A vSphere feature that manages storage latency and allocation among virtual machines.
vSphere HA heartbeat datastore: Datastore storage selected for High Availability heartbeat files.
Storage DRS: A vSphere feature that manages placement and balances virtual machine files across datastores in a datastore cluster.

Choosing the Correct Action

Use the following rule: choose delete only when the datastore and its contents can be permanently discarded. Choose unmount when the data must remain on the storage device or when only one ESXi host should lose access.

Retiring an obsolete shared datastore: Delete it after all workloads and required files have been migrated or backed up and the storage is being permanently decommissioned.
Removing one ESXi host from shared storage: Unmount it from that host so other hosts can retain access to the data.
Preparing storage for maintenance: Unmount it when host access must be detached but the VMFS contents must be preserved.
Unmounting a datastore containing a powered-off VM: Do not proceed. Move the VM off the datastore first because unmounting requires that no VMs reside there.

Final confirmation checklist

Have you selected the intended datastore and verified its storage identity?
Do you know whether the datastore is shared by multiple ESXi hosts?
Have all virtual machines and virtual disks been identified?
Are affected virtual machines shut down and migrated or backed up?
Have non-VM files, templates, and ISO files been preserved if needed?
Are you choosing delete only for permanent, destructive removal?
For unmounting, are there no VMs, datastore-cluster membership, Storage I/O Control, HA heartbeat, or Storage DRS dependencies?
Have you reviewed the final confirmation prompt and considered the effect on other hosts?

Troubleshooting Removal Problems

The unmount action is unavailable or fails

Common causes include a virtual machine still residing on the datastore, membership in a datastore cluster, enabled Storage I/O Control, HA heartbeat usage, or Storage DRS usage. Identify each dependency, remove or relocate it, and retry the unmount.

Files are no longer visible after deletion

The destructive delete action may have been selected instead of unmount, or required content may not have been backed up. Treat datastore deletion as destructive and restore required content from an available backup where possible.

A removal attempt could affect running workloads

Virtual machine disks may still be located on the datastore, or the virtual machines using those disks may not have been shut down. Stop the affected workloads and migrate or otherwise remove their datastore dependency before continuing.