VMware

Clarification on disk reclamation on VMFS5 when using thick provisioning

I thought this only applied to thin-provisioning but I have noticed on a VMFS5 datastore, when I delete a VM or ISO’s it does not re-claim the space back.

Am I right in saying that I need to manually claim this back in the same manner as if thin-provisioning was used for the VMs?

If so, is this a simple case of using the following:

esxcli storage vmfs unmap –volume-label=volume_label|–volume-uuid=volume_uuid –reclaim-unit=number

ESXI = v6.0

vCentre = v6.0

Datastores = VMFS5


View Reddit by tclewesView Source

Related Articles

One Comment

  1. yes with vmfs5, if the lun is thin provisioned from the array (most are), then you’ll need to manually unmap the freed blocks. When you delete things, the space will show as freed in vcenter and on the hosts, but not at the array until you unmap the freed blocks.

    ​

    [https://kb.vmware.com/s/article/2057513](https://kb.vmware.com/s/article/2057513)

    ​

    with vmfs6, you can auto-reclaim this space. This is not an option for you being on esxi 6.0. you’ll need to upgrade to at least 6.5.

    [https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.storage.doc/GUID-98901BEE-8675-4D39-8AA2-E790EE341986.html](https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.storage.doc/GUID-98901BEE-8675-4D39-8AA2-E790EE341986.html)

    ​

    ​

    it should be noted that you can’t upgrade a datastore/lun from vmfs5 to vmfs6. you’ll need to provision new luns and then migrate the data over.

Leave a Reply

Your email address will not be published. Required fields are marked *

Close