Reverting a Memory Snapshot Fails with Resuming virtual disk failed error

Products

• VMware vSphere ESXi
• VMware vCenter Server 8.0

Issue/Introduction

• When attempting to revert a virtual machine to a memory-inclusive snapshot, the task fails in the vCenter UI (VM > Monitor > Tasks).
  • The following errors are observed in the vCenter Tasks:
    
    "Module 'CheckpointLate' power on failed."
    
    
  • More importantly, ESXi hostd logs show below log-line: 
    
    "Resuming virtual disk <REDACTED_HOSTNAMES> failed. The disk has been modified since a snapshot was taken or the virtual machine was suspended."
    
    
• If the snapshot is taken by Not selecting 'include memory' , then the revert is successful.

Environment

• VMware vSphere 8.x
• vCenter Server 8.x

Cause

The issue occurs because one or more virtual disks attached to the VM (e.g., scsi0:1) are configured with the independent-persistent disk mode.

Independent-persistent disks are excluded from the snapshot mechanism and do not utilize delta/redo files, meaning they continue to accept direct writes after a snapshot is taken. When a memory-inclusive snapshot is reverted, the checkpoint restore process attempts to resume the VM from the saved memory state. This saved memory state references disk blocks exactly as they existed at the moment the snapshot was created. Because the independent disk was modified after the snapshot was taken, the hypervisor detects a critical mismatch between the current on-disk state and the expected in-memory state, resulting in the CheckpointLate failure.

Resolution

Determine if the independent-persistent disk configuration is intentional (e.g., configured specifically for a database transaction log that should not be rolled back).

Scenario A: The independent-persistent configuration is intentional If the disk must remain in independent-persistent mode, memory-inclusive snapshots are not supported for this virtual machine.

1. Cease using the "Snapshot the virtual machine's memory" option.

2. Take standard snapshots with the memory option deselected. Non-memory reverts will succeed because they boot the VM fresh and do not require reconciling saved memory states with modified disks.

Scenario B: The independent-persistent configuration is accidental If the disk was incorrectly configured and can be safely changed to dependent mode, perform the following steps to resolve the issue:

1. Ensure the key provider is accessible and functioning (required only if the VM is encrypted).

2. Create a backup of the VM or a copy of the independent-persistent VMDK before proceeding.

3. Power off the virtual machine.

4. Consolidate and delete ALL existing snapshots. Changing disk modes with active snapshots will cause metadata inconsistencies in the snapshot chain, as the .vmsd file currently tracks the disk under mode 1 (independent-persistent).

5. Edit the virtual machine settings and change the target disk mode from independent-persistent to dependent.

6. Power on the virtual machine.

7. Take a new memory-inclusive snapshot and attempt to revert to validate the fix.

Additional Information

For other problems related to snapshot revert failures, check the below KBs:

Powering on a virtual machine from a suspend state or reverting to a snapshot fails

Failed to Power ON Virtual Machine