2.8. Virtual Disks

2.8.1. Understanding Virtual Machine Storage

Red Hat Virtualization supports three storage types: NFS, iSCSI and FCP.

In each type, a host known as the Storage Pool Manager (SPM) manages access between hosts and storage. The SPM host is the only node that has full access within the storage pool; the SPM can modify the storage domain metadata, and the pool’s metadata. All other hosts can only access virtual machine hard disk image data.

By default in an NFS, local, or POSIX compliant data center, the SPM creates the virtual disk using a thin provisioned format as a file in a file system.

In iSCSI and other block-based data centers, the SPM creates a volume group on top of the Logical Unit Numbers (LUNs) provided, and makes logical volumes to use as virtual disks. Virtual disks on block-based storage are preallocated by default.

If the virtual disk is preallocated, a logical volume of the specified size in GB is created. The virtual machine can be mounted on a Red Hat Enterprise Linux server using kpartx, vgscan, vgchange or mount to investigate the virtual machine’s processes or problems.

If the virtual disk is thinly provisioned, a 1 GB logical volume is created. The logical volume is continuously monitored by the host on which the virtual machine is running. As soon as the usage nears a threshold the host notifies the SPM, and the SPM extends the logical volume by 1 GB. The host is responsible for resuming the virtual machine after the logical volume has been extended. If the virtual machine goes into a paused state it means that the SPM could not extend the disk in time. This occurs if the SPM is too busy or if there is not enough storage space.

A virtual disk with a preallocated (raw) format has significantly faster write speeds than a virtual disk with a thin provisioning (QCOW2) format. Thin provisioning takes significantly less time to create a virtual disk. The thin provision format is suitable for non-I/O intensive virtual machines. The preallocated format is recommended for virtual machines with high I/O writes. If a virtual machine is able to write more than 1 GB every four seconds, use preallocated disks where possible.

2.8.2. Understanding Virtual Disks

Red Hat Virtualization features Preallocated (thick provisioned) and Sparse (thin provisioned) storage options.

  • Preallocated

    A preallocated virtual disk allocates all the storage required for a virtual machine up front. For example, a 20 GB preallocated logical volume created for the data partition of a virtual machine will take up 20 GB of storage space immediately upon creation.

  • Sparse

    A sparse allocation allows an administrator to define the total storage to be assigned to the virtual machine, but the storage is only allocated when required.

    For example, a 20 GB thin provisioned logical volume would take up 0 GB of storage space when first created. When the operating system is installed it may take up the size of the installed file, and would continue to grow as data is added up to a maximum of 20 GB size.

You can view a virtual disk’s ID in StorageDisks. The ID is used to identify a virtual disk because its device name (for example, /dev/vda0) can change, causing disk corruption. You can also view a virtual disk’s ID in /dev/disk/by-id.

You can view the Virtual Size of a disk in StorageDisks and in the Disks tab of the details view for storage domains, virtual machines, and templates. The Virtual Size is the total amount of disk space that the virtual machine can use. It is the number that you enter in the Size(GB) field when you create or edit a virtual disk.

You can view the Actual Size of a disk in the Disks tab of the details view for storage domains and templates. This is the amount of disk space that has been allocated to the virtual machine so far. Preallocated disks show the same value for Virtual Size and Actual Size. Sparse disks may show different values, depending on how much disk space has been allocated.

The possible combinations of storage types and formats are described in the following table.

Table 2.35. Permitted Storage Combinations

StorageFormatTypeNote

NFS

Raw

Preallocated

A file with an initial size that equals the amount of storage defined for the virtual disk, and has no formatting.

NFS

Raw

Sparse

A file with an initial size that is close to zero, and has no formatting.

NFS

QCOW2

Sparse

A file with an initial size that is close to zero, and has QCOW2 formatting. Subsequent layers will be QCOW2 formatted.

SAN

Raw

Preallocated

A block device with an initial size that equals the amount of storage defined for the virtual disk, and has no formatting.

SAN

QCOW2

Sparse

A block device with an initial size that is much smaller than the size defined for the virtual disk (currently 1 GB), and has QCOW2 formatting for which space is allocated as needed (currently in 1 GB increments).

2.8.3. Settings to Wipe Virtual Disks After Deletion

The wipe_after_delete flag, viewed in the Administration Portal as the Wipe After Delete check box will replace used data with zeros when a virtual disk is deleted. If it is set to false, which is the default, deleting the disk will open up those blocks for reuse but will not wipe the data. It is, therefore, possible for this data to be recovered because the blocks have not been returned to zero.

The wipe_after_delete flag only works on block storage. On file storage, for example NFS, the option does nothing because the file system will ensure that no data exists.

Enabling wipe_after_delete for virtual disks is more secure, and is recommended if the virtual disk has contained any sensitive data. This is a more intensive operation and users may experience degradation in performance and prolonged delete times.

Note

The wipe after delete functionality is not the same as secure delete, and cannot guarantee that the data is removed from the storage, just that new disks created on same storage will not expose data from old disks.

The wipe_after_delete flag default can be changed to true during the setup process (see Configuring the Red Hat Virtualization Manager), or by using the engine-config tool on the Red Hat Virtualization Manager. Restart the ovirt-engine service for the setting change to take effect.

Note

Changing the wipe_after_delete flag’s default setting will not affect the Wipe After Delete property of disks that already exist.

Setting SANWipeAfterDelete to Default to True Using the Engine Configuration Tool

  1. Run the engine-config tool with the --set action:

    # engine-config --set SANWipeAfterDelete=true
  2. Restart the ovirt-engine service for the change to take effect:

    # systemctl restart ovirt-engine.service

The /var/log/vdsm/vdsm.log file located on the host can be checked to confirm that a virtual disk was successfully wiped and deleted.

For a successful wipe, the log file will contain the entry, storage_domain_id/volume_id was zeroed and will be deleted. For example:

a9cb0625-d5dc-49ab-8ad1-72722e82b0bf/a49351a7-15d8-4932-8d67-512a369f9d61 was zeroed and will be deleted

For a successful deletion, the log file will contain the entry, finished with VG:storage_domain_id LVs: list_of_volume_ids, img: image_id. For example:

finished with VG:a9cb0625-d5dc-49ab-8ad1-72722e82b0bf LVs: {'a49351a7-15d8-4932-8d67-512a369f9d61': ImgsPar(imgs=['11f8b3be-fa96-4f6a-bb83-14c9b12b6e0d'], parent='00000000-0000-0000-0000-000000000000')}, img: 11f8b3be-fa96-4f6a-bb83-14c9b12b6e0d

An unsuccessful wipe will display a log message zeroing storage_domain_id/volume_id failed. Zero and remove this volume manually, and an unsuccessful delete will display Remove failed for some of VG: storage_domain_id zeroed volumes: list_of_volume_ids.

2.8.4. Shareable Disks in Red Hat Virtualization

Some applications require storage to be shared between servers. Red Hat Virtualization allows you to mark virtual machine hard disks as Shareable and attach those disks to virtual machines. That way a single virtual disk can be used by multiple cluster-aware guests.

Shared disks are not to be used in every situation. For applications like clustered database servers, and other highly available services, shared disks are appropriate. Attaching a shared disk to multiple guests that are not cluster-aware is likely to cause data corruption because their reads and writes to the disk are not coordinated.

You cannot take a snapshot of a shared disk. Virtual disks that have snapshots taken of them cannot later be marked shareable.

You can mark a disk shareable either when you create it, or by editing the disk later.

Important

Only RAW format disks can be made shareable.

2.8.5. Read Only Disks in Red Hat Virtualization

Some applications require administrators to share data with read-only rights. You can do this when creating or editing a disk attached to a virtual machine via the Disks tab in the details view of the virtual machine and selecting the Read Only check box. That way, a single disk can be read by multiple cluster-aware guests, while an administrator maintains writing privileges.

You cannot change the read-only status of a disk while the virtual machine is running.

Important

Mounting a journaled file system requires read-write access. Using the Read Only option is not appropriate for virtual disks that contain such file systems (e.g. EXT3, EXT4, or XFS).

2.8.6. Virtual Disk Tasks

2.8.6.1. Creating a Virtual Disk

Image disk creation is managed entirely by the Manager. Direct LUN disks require externally prepared targets that already exist.

You can create a virtual disk that is attached to a specific virtual machine. Additional options are available when creating an attached virtual disk, as specified in Explanation of Settings in the New Virtual Disk Window.

Creating a Virtual Disk Attached to a Virtual Machine

  1. Click ComputeVirtual Machines.
  2. Click the virtual machine’s name. This opens the details view.
  3. Click the Disks tab.
  4. Click New.
  5. Click the appropriate button to specify whether the virtual disk will be an Image or Direct LUN disk.
  6. Select the options required for your virtual disk. The options change based on the disk type selected. See Explanation of Settings in the New Virtual Disk Window for more details on each option for each disk type.
  7. Click OK.

You can also create a floating virtual disk that does not belong to any virtual machines. You can attach this disk to a single virtual machine, or to multiple virtual machines if the disk is shareable. Some options are not available when creating a virtual disk, as specified in Explanation of Settings in the New Virtual Disk Window.

Creating a Floating Virtual Disk

  1. Click StorageDisks.
  2. Click New.
  3. Click the appropriate button to specify whether the virtual disk will be an Image or Direct LUN disk.
  4. Select the options required for your virtual disk. The options change based on the disk type selected. See Explanation of Settings in the New Virtual Disk Window for more details on each option for each disk type.
  5. Click OK.

2.8.6.2. Explanation of settings in the New Virtual Disk window

Because the New Virtual Disk windows for creating floating and attached virtual disks are very similar, their settings are described in a single section.

Table 2.36. New Virtual Disk and Edit Virtual Disk settings: Image

Field NameDescription

Size(GB)

The size of the new virtual disk in GB.

Alias

The name of the virtual disk, limited to 40 characters.

Description

A description of the virtual disk. This field is recommended but not mandatory.

Interface

This field only appears when creating an attached disk.

The virtual interface the disk presents to virtual machines. VirtIO is faster, but requires drivers. Red Hat Enterprise Linux 5 and later include these drivers. Windows does not include these drivers, but you can install them from the virtio-win ISO image. IDE and SATA devices do not require special drivers.

The interface type can be updated after stopping all virtual machines that the disk is attached to.

Data Center

This field only appears when creating a floating disk.

The data center in which the virtual disk will be available.

Storage Domain

The storage domain in which the virtual disk will be stored. The drop-down list shows all storage domains available in the given data center, and also shows the total space and currently available space in the storage domain.

Allocation Policy

The provisioning policy for the new virtual disk.

  • Preallocated allocates the entire size of the disk on the storage domain at the time the virtual disk is created. The virtual size and the actual size of a preallocated disk are the same. Preallocated virtual disks take more time to create than thin provisioned virtual disks, but have better read and write performance. Preallocated virtual disks are recommended for servers and other I/O intensive virtual machines. If a virtual machine is able to write more than 1 GB every four seconds, use preallocated disks where possible.
  • Thin Provision allocates 1 GB at the time the virtual disk is created and sets a maximum limit on the size to which the disk can grow. The virtual size of the disk is the maximum limit; the actual size of the disk is the space that has been allocated so far. Thin provisioned disks are faster to create than preallocated disks and allow for storage over-commitment. Thin provisioned virtual disks are recommended for desktops.

Disk Profile

The disk profile assigned to the virtual disk. Disk profiles define the maximum amount of throughput and the maximum level of input and output operations for a virtual disk in a storage domain. Disk profiles are defined on the storage domain level based on storage quality of service entries created for data centers.

Activate Disk(s)

This field only appears when creating an attached disk.

Activate the virtual disk immediately after creation.

Wipe After Delete

Allows you to enable enhanced security for deletion of sensitive material when the virtual disk is deleted.

Bootable

This field only appears when creating an attached disk.

Allows you to enable the bootable flag on the virtual disk.

Shareable

Allows you to attach the virtual disk to more than one virtual machine at a time.

Read-Only

This field only appears when creating an attached disk.

Allows you to set the disk as read-only. The same disk can be attached as read-only to one virtual machine, and as rewritable to another.

Enable Incremental Backup

Enables incremental backup on the virtual disk. Incremental backup requires disks to be formatted in QCOW2 format instead of RAW format. See Incremental backup and restore.

Enable Discard

This field only appears when creating an attached disk.

Allows you to shrink a thin provisioned disk while the virtual machine is up. For block storage, the underlying storage device must support discard calls, and the option cannot be used with Wipe After Delete unless the underlying storage supports the discard_zeroes_data property. For file storage, the underlying file system and the block device must support discard calls. If all requirements are met, SCSI UNMAP commands issued from guest virtual machines is passed on by QEMU to the underlying storage to free up the unused space.

The Direct LUN settings can be displayed in either Targets > LUNs or LUNs > Targets. Targets > LUNs sorts available LUNs according to the host on which they are discovered, whereas LUNs > Targets displays a single list of LUNs.

Fill in the fields in the Discover Targets section and click Discover to discover the target server. You can then click the Login All button to list the available LUNs on the target server and, using the radio buttons next to each LUN, select the LUN to add.

Using LUNs directly as virtual machine hard disk images removes a layer of abstraction between your virtual machines and their data.

The following considerations must be made when using a direct LUN as a virtual machine hard disk image:

  • Live storage migration of direct LUN hard disk images is not supported.
  • Direct LUN disks are not included in virtual machine exports.
  • Direct LUN disks are not included in virtual machine snapshots.

Table 2.37. New Virtual Disk and Edit Virtual Disk settings: Direct LUN

Field NameDescription

Alias

The name of the virtual disk, limited to 40 characters.

Description

A description of the virtual disk. This field is recommended but not mandatory. By default the last 4 characters of the LUN ID is inserted into the field.

The default behavior can be configured by setting the PopulateDirectLUNDiskDescriptionWithLUNId configuration key to the appropriate value using the engine-config command. The configuration key can be set to -1 for the full LUN ID to be used, or 0 for this feature to be ignored. A positive integer populates the description with the corresponding number of characters of the LUN ID.

Interface

This field only appears when creating an attached disk.

The virtual interface the disk presents to virtual machines. VirtIO is faster, but requires drivers. Red Hat Enterprise Linux 5 and later include these drivers. Windows does not include these drivers, but they can be installed from the virtio-win ISO . IDE and SATA devices do not require special drivers.

The interface type can be updated after stopping all virtual machines that the disk is attached to.

Data Center

This field only appears when creating a floating disk.

The data center in which the virtual disk will be available.

Host

The host on which the LUN will be mounted. You can select any host in the data center.

Storage Type

The type of external LUN to add. You can select from either iSCSI or Fibre Channel.

Discover Targets

This section can be expanded when you are using iSCSI external LUNs and Targets > LUNs is selected.

Address - The host name or IP address of the target server.

Port - The port by which to attempt a connection to the target server. The default port is 3260.

User Authentication - The iSCSI server requires User Authentication. The User Authentication field is visible when you are using iSCSI external LUNs.

CHAP user name - The user name of a user with permission to log in to LUNs. This field is accessible when the User Authentication check box is selected.

CHAP password - The password of a user with permission to log in to LUNs. This field is accessible when the User Authentication check box is selected.

Activate Disk(s)

This field only appears when creating an attached disk.

Activate the virtual disk immediately after creation.

Bootable

This field only appears when creating an attached disk.

Allows you to enable the bootable flag on the virtual disk.

Shareable

Allows you to attach the virtual disk to more than one virtual machine at a time.

Read-Only

This field only appears when creating an attached disk.

Allows you to set the disk as read-only. The same disk can be attached as read-only to one virtual machine, and as rewritable to another.

Enable Discard

This field only appears when creating an attached disk.

Allows you to shrink a thin provisioned disk while the virtual machine is up. With this option enabled, SCSI UNMAP commands issued from guest virtual machines is passed on by QEMU to the underlying storage to free up the unused space.

Enable SCSI Pass-Through

This field only appears when creating an attached disk.

Available when the Interface is set to VirtIO-SCSI. Selecting this check box enables passthrough of a physical SCSI device to the virtual disk. A VirtIO-SCSI interface with SCSI passthrough enabled automatically includes SCSI discard support. Read-Only is not supported when this check box is selected.

When this check box is not selected, the virtual disk uses an emulated SCSI device. Read-Only is supported on emulated VirtIO-SCSI disks.

Allow Privileged SCSI I/O

This field only appears when creating an attached disk.

Available when the Enable SCSI Pass-Through check box is selected. Selecting this check box enables unfiltered SCSI Generic I/O (SG_IO) access, allowing privileged SG_IO commands on the disk. This is required for persistent reservations.

Using SCSI Reservation

This field only appears when creating an attached disk.

Available when the Enable SCSI Pass-Through and Allow Privileged SCSI I/O check boxes are selected. Selecting this check box disables migration for any virtual machine using this disk, to prevent virtual machines that are using SCSI reservation from losing access to the disk.

Important

Mounting a journaled file system requires read-write access. Using the Read-Only option is not appropriate for virtual disks that contain such file systems (e.g. EXT3, EXT4, or XFS).

2.8.6.3. Overview of Live Storage Migration

Virtual disks can be migrated from one storage domain to another while the virtual machine to which they are attached is running. This is referred to as live storage migration. When a disk attached to a running virtual machine is migrated, a snapshot of that disk’s image chain is created in the source storage domain, and the entire image chain is replicated in the destination storage domain. As such, ensure that you have sufficient storage space in both the source storage domain and the destination storage domain to host both the disk image chain and the snapshot. A new snapshot is created on each live storage migration attempt, even when the migration fails.

Consider the following when using live storage migration:

  • You can live migrate multiple disks at one time.
  • Multiple disks for the same virtual machine can reside across more than one storage domain, but the image chain for each disk must reside on a single storage domain.
  • You can live migrate disks between any two storage domains in the same data center.
  • You cannot live migrate direct LUN hard disk images or disks marked as shareable.

2.8.6.4. Moving a Virtual Disk

Move a virtual disk that is attached to a virtual machine or acts as a floating virtual disk from one storage domain to another. You can move a virtual disk that is attached to a running virtual machine; this is referred to as live storage migration. Alternatively, shut down the virtual machine before continuing.

Consider the following when moving a disk:

  • You can move multiple disks at the same time.
  • You can move disks between any two storage domains in the same data center.
  • If the virtual disk is attached to a virtual machine that was created based on a template and used the thin provisioning storage allocation option, you must copy the disks for the template on which the virtual machine was based to the same storage domain as the virtual disk.

Procedure

  1. Click StorageDisks and select one or more virtual disks to move.
  2. Click Move.
  3. From the Target list, select the storage domain to which the virtual disk(s) will be moved.
  4. From the Disk Profile list, select a profile for the disk(s), if applicable.
  5. Click OK.

The virtual disks are moved to the target storage domain. During the move procedure, the Status column displays Locked and a progress bar indicating the progress of the move operation.

2.8.6.5. Changing the Disk Interface Type

Users can change a disk’s interface type after the disk has been created. This enables you to attach an existing disk to a virtual machine that requires a different interface type. For example, a disk using the VirtIO interface can be attached to a virtual machine requiring the VirtIO-SCSI or IDE interface. This provides flexibility to migrate disks for the purpose of backup and restore, or disaster recovery. The disk interface for shareable disks can also be updated per virtual machine. This means that each virtual machine that uses the shared disk can use a different interface type.

To update a disk interface type, all virtual machines using the disk must first be stopped.

Changing a Disk Interface Type*

  1. Click ComputeVirtual Machines and stop the appropriate virtual machine(s).
  2. Click the virtual machine’s name. This opens the details view.
  3. Click the Disks tab and select the disk.
  4. Click Edit.
  5. From the Interface list, select the new interface type and click OK.

You can attach a disk to a different virtual machine that requires a different interface type.

Attaching a Disk to a Different Virtual Machine using a Different Interface Type

  1. Click ComputeVirtual Machines and stop the appropriate virtual machine(s).
  2. Click the virtual machine’s name. This opens the details view.
  3. Click the Disks tab and select the disk.
  4. Click Remove, then click OK.
  5. Go back to Virtual Machines and click the name of the new virtual machine that the disk will be attached to.
  6. Click the Disks tab, then click Attach.
  7. Select the disk in the Attach Virtual Disks window and select the appropriate interface from the Interface drop-down.
  8. Click OK.

2.8.6.6. Copying a Virtual Disk

You can copy a virtual disk from one storage domain to another. The copied disk can be attached to virtual machines.

Procedure

  1. Click StorageDisks and select the virtual disk(s).
  2. Click Copy .
  3. Optionally, enter a new name in the Alias field.
  4. From the Target list, select the storage domain to which the virtual disk(s) will be copied.
  5. From the Disk Profile list, select a profile for the disk(s), if applicable.
  6. Click OK.

The virtual disks have a status of Locked while being copied.

2.8.6.7. Improving disk performance

In the Administration Portal, on the virtual machine’s Resource Allocation tab, the default I/O Threads Enabled setting is checked (enabled), and the number of threads is 1.

Suppose a virtual machine has multiple disks that have VirtIO controllers, and its workloads make significant use of those controllers. In that case, you can improve performance by increasing the number of I/O threads.

However, also consider that increasing the number of I/O threads decreases the virtual machine’s pool of threads. If your workloads do not use the VirtIO controllers and the threads you allocate to them, increasing the number of I/O threads might decrease overall performance.

To find the optimal number of threads, benchmark the performance of your virtual machine running workloads before and after you adjust the number of threads.

Procedure

  1. On ComputeVirtual Machines, Power Off the virtual machine.
  2. Click the name of the virtual machine.
  3. In the details pane, click the Vm Devices tab.
  4. Count the number of controllers whose Type is virtio or virtio-scsi.
  5. Click Edit.
  6. In the Edit Virtual Machine window, click the Resource Allocation tab.
  7. Confirm that I/O Threads Enabled is checked (enabled).
  8. To the right of I/O Threads Enabled, increase the number of threads, but do not exceed number of controllers whose type is virtio or virtio-scsi.
  9. Click OK.
  10. In the details pane, click the Disks tab.
  11. For each disk, use More Actions ( moreactions ) to Deactivate and Activate the disk. This action remaps the disks to the controllers.
  12. Click Run to start the virtual machine.

Verification steps

  • To see which controllers have an I/O thread, click Vm Devices in the details pane and look for ioThreadid= in the Spec Params column.
  • To see the mapping of disks to controllers, log into the host machine and enter the following command:

    # virsh -r dumpxml virtual_machine_name

2.8.6.8. Uploading Images to a Data Storage Domain

You can upload virtual disk images and ISO images to your data storage domain in the Administration Portal or with the REST API. See Uploading Images to a Data Storage Domain for details.

2.8.6.9. Importing a Disk Image from an Imported Storage Domain

Import floating virtual disks from an imported storage domain.

Note

Only QEMU-compatible disks can be imported into the Manager.

Procedure

  1. Click StorageDomains.
  2. Click the name of an imported storage domain. This opens the details view.
  3. Click the Disk Import tab.
  4. Select one or more disks and click Import.
  5. Select the appropriate Disk Profile for each disk.
  6. Click OK.

2.8.6.10. Importing an Unregistered Disk Image from an Imported Storage Domain

Import floating virtual disks from a storage domain. Floating disks created outside of a Red Hat Virtualization environment are not registered with the Manager. Scan the storage domain to identify unregistered floating disks to be imported.

Note

Only QEMU-compatible disks can be imported into the Manager.

Procedure

  1. Click StorageDomains.
  2. Click the storage domain’s name. This opens the details view.
  3. Click More Actions ( moreactions ), then click Scan Disks so that the Manager can identify unregistered disks.
  4. Click the Disk Import tab.
  5. Select one or more disk images and click Import.
  6. Select the appropriate Disk Profile for each disk.
  7. Click OK.

2.8.6.11. Importing a Virtual Disk from an OpenStack Image Service

Virtual disks managed by an OpenStack Image Service can be imported into the Red Hat Virtualization Manager if that OpenStack Image Service has been added to the Manager as an external provider.

  1. Click StorageDomains.
  2. Click the OpenStack Image Service domain’s name. This opens the details view.
  3. Click the Images tab and select an image.
  4. Click Import.
  5. Select the Data Center into which the image will be imported.
  6. From the Domain Name drop-down list, select the storage domain in which the image will be stored.
  7. Optionally, select a quota to apply to the image from the Quota drop-down list.
  8. Click OK.

The disk can now be attached to a virtual machine.

2.8.6.12. Exporting a Virtual Disk to an OpenStack Image Service

Virtual disks can be exported to an OpenStack Image Service that has been added to the Manager as an external provider.

Important

Virtual disks can only be exported if they do not have multiple volumes, are not thin provisioned, and do not have any snapshots.

  1. Click StorageDisks and select the disks to export.
  2. Click More Actions ( moreactions ), then click Export.
  3. From the Domain Name drop-down list, select the OpenStack Image Service to which the disks will be exported.
  4. From the Quota drop-down list, select a quota for the disks if a quota is to be applied.
  5. Click OK.

2.8.6.13. Reclaiming Virtual Disk Space

Virtual disks that use thin provisioning do not automatically shrink after deleting files from them. For example, if the actual disk size is 100GB and you delete 50GB of files, the allocated disk size remains at 100GB, and the remaining 50GB is not returned to the host, and therefore cannot be used by other virtual machines. This unused disk space can be reclaimed by the host by performing a sparsify operation on the virtual machine’s disks. This transfers the free space from the disk image to the host. You can sparsify multiple virtual disks in parallel.

Perform this operation before cloning a virtual machine, creating a template based on a virtual machine, or cleaning up a storage domain’s disk space.

Limitations

  • NFS storage domains must use NFS version 4.2 or higher.
  • You cannot sparsify a disk that uses a direct LUN.
  • You cannot sparsify a disk that uses a preallocated allocation policy. If you are creating a virtual machine from a template, you must select Thin from the Storage Allocation field, or if selecting Clone, ensure that the template is based on a virtual machine that has thin provisioning.
  • You can only sparsify active snapshots.

Sparsifying a Disk

  1. Click ComputeVirtual Machines and shut down the required virtual machine.
  2. Click the virtual machine’s name. This opens the details view.
  3. Click the Disks tab. Ensure that the disk’s status is OK.
  4. Click More Actions ( moreactions ), then click Sparsify.
  5. Click OK.

A Started to sparsify event appears in the Events tab during the sparsify operation and the disk’s status displays as Locked. When the operation is complete, a Sparsified successfully event appears in the Events tab and the disk’s status displays as OK. The unused disk space has been returned to the host and is available for use by other virtual machines.