I wonder what holds back a ZFS-level offline dedupe function now that that's implemented since you could already basically write a shell script to do something like it.
I tried it a few months ago and ReFS ate my data. No indication of why in event logs or SMART data. It had IsPowerProtected set because I have a UPS and I had a unclean restart, I would expect it to lose data, but not to corrupt the filesystem metadata. I had a backup of the data but wanted some recent changes. Refsutil (the official Microsoft tool) didn't help because it has not been updated for the newest ReFS version. I couldn't read most files because I had integrity enable and files failed the check. Hetman's Data Recovery was able to recover most of the data. In later testing I found out that IsPowerProtected is just very unsafe. I have since put some time into testing and sometimes fixing https://github.com/openzfsonwindows/openzfs , it is not ready for use yet, but it is making great progress.
Might be easy to extende fdupes and jdupes to be able to do this without much effort. I haven't seen the api/syscall invokved but I use them with btrfs for a specific use where I have a lot of known duplicates.
I wonder what holds back a ZFS-level offline dedupe function now that that's implemented since you could already basically write a shell script to do something like it.