PowerVault ME4, ME5: Host operating system space consumption does not match pool space consumption

• Capacity measurement units
• White space recovery on thin provisioned volumes

Capacity measurement units

By default PowerVault ME users are configured to report volume size and pool capacity consumption using base 10 measurement units, whereas host operating systems use base 2 measurement units to report LUN capacity or file system space consumption. Administrators may wonder why a volume created in PowerVault Manager as 140 TB capacity is presented as a LUN to the host operating system measuring 128 TB (TiB).

Most host operating systems use base 2 as the unit of measurement, and notate this as TB rather than TiB leading to confusion for administrators unfamiliar with the measurement unit conventions used. On PowerVault ME series arrays the measurement units displayed are set per user.

• PowerVault ME4 series
• PowerVault ME5 series

PowerVault ME4 series arrays changing measurement units

1. Log in as a user with the manage role and perform one of the following:

   • In the Home topic, select Action > System Settings, then click the Managing Users tab.
   • In the System topic, select Action > System Settings, then click the Manage Users tab.
   • In the banner, click the user panel and select Manage Users.
   • In the Welcome panel, select System Settings > Manage Users. The Manage Users tab displays a table of existing users and options to set.

2. Select the user to modify.

3. Change the settings. You cannot change the username. Users with the monitor role can change their own settings except for their role and interface settings.

4. Perform one of the following:

   • To save your settings and continue configuring your system, click Apply.
   • To save your settings and close the panel, click Apply and Close.

   A confirmation panel appears.

5. Click OK to save your changes. Otherwise, click Cancel.

PowerVault ME5 series arrays changing management units
This can be changed using command-line interface (CLI), follow the instructions outlined in PowerVault ME5: No Base unit preference option available while creating a user through PowerVault Manager

White space recovery on thin provisioned volumes

PowerVault ME series can be deployed in virtual pool mode (default recommended) or linear pool mode. Virtual pool mode supports snapshots, replication, tiering and over-provisioning volumes. All volumes created in virtual pools are thin provisioned, meaning physical space in the pool is consumed as the file system writes to volumes.

Overcommitting (over-provisioning) is a virtual storage feature that allows a system administrator to overcommit physical storage resources. This allows the host system to operate as though it has more storage available than is actually allocated to it. When physical resources fill up, the administrator can add physical storage by adding additional disk groups or expanding an ADAPT disk group. This feature can be disabled by clearing the option in pool settings. See the PowerVault ME administrator guide or PowerVault Manager online help for instructions to change this setting.

PowerVault ME series arrays are block storage, as host applications write to the file system new blocks are assigned to the volume from the pool, except when overwriting existing blocks previously used by the file system. When files are deleted from the host operating file system, the blocks remain assigned to the volume. Administrators may report that the free space reported by the host file system does not match the pool free space (blocks), the term "whitespace" is often used to describe blocks which are allocated on the storage LUN/Volume but not in use by the filesystem.

The SCSI UNMAP command is issued by host operating systems to reclaim unused storage space in thin-provisioned storage environments. How this command is implemented varies by host operating system and file system used. Administrators must consult their host operating system documentation for specific instructions. Options for the most common operating systems and file systems are outlined below.

Microsoft Windows Server
SCSI UNMAP support is enabled for NTFS file systems by default. System administrators can use the Windows optimize tool. For systems that use REFS file system consult Microsoft documentation for information about restrictions and limitations.

Administrators can review the following Microsoft links to learn more.

• Thin Provisioning - Windows drivers | Microsoft Learn
• Check the disabledeletenotify setting using fsutil behavior | Microsoft Learn
• REFS users may use SDelete - Sysinternals | Microsoft Learn
• PowerShell users can use Optimize-Volume (Storage) | Microsoft Learn

Broadcom VMware ESXi
PowerVault ME virtual volumes use 4 MB page size, there is no automatic asynchronous reclamation of free space on VMFS6 datastores in ESXi. ESXi administrators must run the command esxcli storage vmfs unmap to reclaim unused blocks.

esxcli storage vmfs unmap --volume-label=volume_label

For more details, see Storage Space Reclamation (vmware.com)

Linux
Linux users can mount host file systems with the discard option and use the fstrim command to recover whitespace. SCSI UNMAP support varies by kernel version, and file system used see vendors documentation for recommendations or limitations.

For Red Hat Linux distributions, see Discarding unused blocks | Red Hat Product Documentation

For SuSE enterprise Linux , see Disabling fstrim - under which conditions? | Support | SUSE