Why does deleting a file in the live system affect snapraid-btrfs?

[FlexibleToast]

As I test I have setup a OMV6 machine in a VM with 10 disks that are pooled with mergerfs and 2 more disks for parity (total of 12 disks). This mirrors my NAS in architecture. I used a for loop to create a bunch of files, similar to:

for i in {01..50}; do
  dd if=/dev/random of=/dev/mergedfs/backend/testfile-${i} bs=1G count=1
done

I then did a sync of all the disks snapraid-btrfs sync. When the sync was done I confirmed that snapper successfully created the snapshots with snapper ls -a. I then deleted a portion of those files that I knew would delete files from multiple drives rm -f /dev/mergedfs/backend/testfile-2*. Running a snapraid-btrfs fix then did what you would expect and read from a temp config. However, several files were unrecoverable. Why would they be unrecoverable if it's supposed to be reading/building parity from the snapshots that still exists?