When you try to back up guest virtual machines (VMs) on your host Hyper-VM computer in Windows Server 2012 R2, you receive one of the following error messages at the start of the operation:
Error message 1
The autorecovery operation failed to complete on the shadow copy. // #define VSS_E_AUTORECOVERY_FAILED ((HRESULT)0x800423FBL)
Error message 2
<VMName> could not initiate a checkpoint operation: The file exists. (0x80070050).
This problem occurs because of a failed or incomplete backup that did not correctly include the AutoRecovery.avhdx file.
During a Volume Shadow Copy Service (VSS) snapshot of volumes for a VM, the writers for the backup flush out the outstanding transactions. The snapshot is created by using this information, and is then set to archive. This snapshot is saved as the AutoRecovery.avhdx file. Now, you have forked a timeline between the flush transactions and the previous state of that VM from before the flush of transactions. This includes the snapshot and any changes that are made to the snapshot after the snapshot is taken.
You can think of the snapshot as having the following differencing disks:
The running VM
The Autorecovery.avhdx file
Microsoft Virtual Machine Manager (VMM) switches the VM to run on the first differencing disk while the VSS snapshot is active. Then, VMM mounts the AutoRecovery.avhdx file to complete the post-snapshot process and remove the AutoRecovery.avhdx file.
To work around this problem, delete or rename the Autorecovery.avhdx file.
Warning This method can have unintended consequences if a backup operation is in progress.
Note As of the publication of this article, you can install and test Microsoft Hyper-V Server 2016 Technical Preview as an alternative workaround to determine whether the problem persists.
Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.