How do I safely deal with a full drive and leftover AVHDX files?

First, don’t delete or rename any AVHDX files. One of them may be the VM’s current disk, and deleting a parent in the chain can make the whole guest unusable. Pause backup or checkpoint jobs that might still be working on this VM, then make a copy of the VM files to other storage if you can.

Free space by moving unrelated data or adding storage, not by guessing which VM disk file is safe to remove. Once there’s room for a merge, check the VM’s disk settings and checkpoint list, and use Hyper-V’s supported merge or checkpoint-removal options. If the chain is unclear or you don’t have a usable backup, get support before changing it.

Check the backup software before trying to clean up anything manually. A backup job can create a temporary checkpoint and leave it behind if the job stalls or fails. Look for an active or stuck job, cancel it only through the backup tool, and check its job history for errors. Then see whether Hyper-V starts merging the disks.

If the backup tool no longer recognizes the checkpoint, or this happens again after a clean run, collect the job logs and Hyper-V events and contact the backup vendor. Repeatedly making new checkpoints or deleting files can make recovery harder.

For a careful diagnosis, inspect the disk chain rather than relying only on the checkpoint list. Hyper-V’s disk inspection tools or PowerShell can show each AVHDX parent path. Compare that chain with the disk path actually attached to the VM. An empty checkpoint list does not by itself prove that every AVHDX is unused.

Only merge once you’ve confirmed the chain and have enough free space. Work from the active child toward its parent using Hyper-V’s merge tools, and keep an untouched copy of the files until the VM boots and the data checks out. If you can’t establish the right order, stop and use a verified backup or get an experienced Hyper-V admin involved.