Skip to content

How to Configure Production Checkpoints in Hyper-V Print

  • 0

Applies to: Windows Server 2019, Windows Server 2022 and Windows Server 2025, subject to the workload and edition requirements below.

Choose production checkpoints with explicit failure behaviour

Production checkpoints create a point-in-time view of supported guest disks using guest coordination rather than capturing the running VM’s memory. Windows guests use VSS; supported Linux guests use file-system freeze mechanisms. Applying a production checkpoint brings the VM back to an offline state that must boot, rather than resuming an old running desktop. That behaviour is usually more suitable for server workloads than a memory snapshot, but application support still matters.

Hyper-V has two similarly named policies with an important difference. Production tries a production checkpoint and can fall back to a standard checkpoint if that attempt fails. ProductionOnly refuses that fallback and fails the request instead. Use ProductionOnly when the operational requirement is to avoid an unnoticed memory-state checkpoint. Neither option makes a checkpoint an independent backup or guarantees consistency across external systems such as another database server.

Confirm guest, workload and storage prerequisites

Verify the guest operating system, integration components and application support for online checkpoints. On a Windows guest, the relevant VSS writers must be healthy and the Hyper-V Volume Shadow Copy Requestor path available. Do not judge health only by a service being set to Automatic: several integration services use trigger-start behaviour. Check the actual integration status and VSS writer results during the intended operation.

Estimate free space on every host volume containing the VM’s disks, configuration and checkpoint files. A checkpoint introduces differencing-disk growth as the VM writes new data. Its storage cost depends on change rate and how long it remains, not simply the size shown immediately after creation. Avoid running a manual checkpoint during a backup job or another checkpoint/merge operation without confirming that the backup product supports that interaction.

Decide what change the checkpoint protects, the acceptance test, the expiry time and who may authorize applying it. Restoring old guest disk state can discard newer work and affect replication partners or external dependencies. Keep a separate recoverable backup before a change with material business consequences.

Inspect and configure the VM policy

Run elevated PowerShell on the Hyper-V host and identify the exact VM. Record the old policy and existing checkpoints before changing anything. The example sets APP01 to ProductionOnly; it does not create a checkpoint until the subsequent command is issued.

Get-VM -Name 'APP01' | Select-Object Name,State,CheckpointType,CheckpointFileLocation
Get-VMCheckpoint -VMName 'APP01'
Get-VMIntegrationService -VMName 'APP01'
Set-VM -Name 'APP01' -CheckpointType ProductionOnly
Get-VM -Name 'APP01' | Select-Object Name,CheckpointType

If the environment is managed through System Center or a backup platform, confirm which system owns the setting. A later policy reconciliation can replace a local choice. Disabling automatic checkpoints, when applicable to the host and VM, is a separate decision from choosing the type used for an explicit checkpoint request. Keep those policies distinct in documentation so a future operator knows what will happen before maintenance.

Validate Windows VSS before creating the checkpoint

Inside a Windows guest, inspect VSS writers and recent application events. Stable writers with no reported error are useful prerequisites, though they are not a guarantee that every application is protected. An application without an appropriate writer may have only the consistency level its own design provides. Follow the application vendor’s requirements for checkpoints and backup.

# Run inside the Windows guest:
vssadmin list writers
Get-Service vmicvss
Get-WinEvent -FilterHashtable @{LogName='Application';StartTime=(Get-Date).AddHours(-1)} -MaxEvents 50

If a writer reports failure, identify its owning application and supported recovery action. Do not restart all VSS-related services or kill processes as a first response on a busy server. Collect the error and time, check whether another backup owns the writer, and coordinate with the workload owner. For Linux, verify the supported guest integration daemons and file-system freeze path rather than applying Windows service commands.

Create a named checkpoint and verify completion

Use a name tied to the actual change and time. The command requests the configured checkpoint type. With ProductionOnly, a failure should remain a visible failure rather than a silent standard-checkpoint fallback. Wait for completion and inspect the resulting checkpoint list and host events.

$name = 'Before-AppUpdate-' + (Get-Date -Format 'yyyyMMdd-HHmm')
Checkpoint-VM -Name 'APP01' -SnapshotName $name -ErrorAction Stop
Get-VMCheckpoint -VMName 'APP01' | Select-Object Name,CreationTime,SnapshotType
Get-VM -Name 'APP01' | Select-Object Name,State,Status
Get-Volume | Select-Object DriveLetter,SizeRemaining,Size

SnapshotType alone is not a substitute for understanding the configured policy and observed creation result. Record the successful request, policy and any VSS or integration warnings. Check that the VM and its application remain healthy before starting the planned change. If creation fails, stop and diagnose; changing to Standard merely to obtain a checkpoint changes the protection semantics and needs a workload-specific decision.

Apply only after accepting the loss of newer state

Applying a checkpoint replaces the VM’s current disk view with the selected older state. It can lose work performed since that point and produce consequences outside the VM, such as mismatched transactions, issued certificates or replicated identities. Confirm the application recovery plan, protect current evidence and coordinate dependent systems before applying it. A generic checkpoint procedure cannot decide that an application rollback is safe.

# Review the exact checkpoint first:
Get-VMCheckpoint -VMName 'APP01' | Select-Object Name,CreationTime
# Recovery action: run only for the selected, approved recovery point:
Restore-VMCheckpoint -VMName 'APP01' -Name $name

Do not suppress confirmation casually. After a production-checkpoint restore, expect to boot the VM and verify the application’s supported recovery behaviour. Check services, data, external dependencies and event logs. A familiar desktop or a Running VM state does not prove that the business transaction state is correct. Keep the post-change evidence until the owner accepts the restored service.

Delete accepted checkpoints through Hyper-V

Once the change is accepted and the checkpoint has served its purpose, remove it through Hyper-V. Deleting the checkpoint consolidates its disk changes into the appropriate parent chain; it does not roll the VM back. Monitor storage and merge completion, particularly on a busy disk or when multiple checkpoints exist.

# Remove only the specifically reviewed checkpoint after acceptance:
Get-VMCheckpoint -VMName 'APP01' -Name $name | Remove-VMCheckpoint
Get-VMCheckpoint -VMName 'APP01'
Get-VMHardDiskDrive -VMName 'APP01'

Never delete AVHDX files manually to reclaim space. They may contain the VM’s current writes, even when a management view is confusing. If a merge fails or storage is nearly exhausted, preserve the chain and collect host, backup and storage evidence. A wrong manual merge or parent-path edit can make data inaccessible. Use the supported recovery procedure for the actual chain and backup state.

Diagnose failures and retain a separate recovery strategy

For production-checkpoint failure, correlate host VMMS events, guest VSS events, integration status, writer health and free space at the same timestamp. Check whether a backup or previous merge is still active. Unsupported disk attachments and workload configurations may prevent checkpoint creation even when an ordinary VHDX-based test VM works. Repeated requests can add complexity without correcting the cause.

Rollback of a policy change means restoring the recorded CheckpointType, not deleting unknown disk files. Keep checkpoints short-lived and monitor their age and growth. Independent backups should survive loss of the host and its storage and should be restored in a controlled test. A production checkpoint is a useful maintenance tool when its consistency and expiry are understood; it is not a replacement for backup retention, application recovery design or a verified path through a host-level disaster.

Related guidance and official references

Create a Generation 2 VM; Configure Hyper-V Replica; Create a server backup.


Was this answer helpful?

« Back