HPE users: if you are using Gen10 platform hardware with Smart Array SR Gen10 RAID controllers for your Veeam repository, please install this firmware update as soon as possible, especially if you use ReFS. The potential data inconsistency the bulletin is talking about appears to be some bug around flush command processing, where "success" is returned by storage before data actually lands on some media. I explained here before how proper flush command handling is essential, because it lets the application persist outstanding I/O to disk before marking the corresponding transaction as completed. For example, this is how we can reliably update our backup files in place. However, if in our case misbehaving flush means corrupted backup files, the impact on modern file systems like ReFS is even worse: with badly timed BSOD, reset or hard power off, this may result in critical file system metadata loss destroying the entire volume. You may have heard stories about ReFS volumes turning raw from Windows perspective – this kind of bugs in RAID controllers is a reason for this issue.