ReFs Drives went to RAW after reinstall of Windows

Exactly what Max said.  I had this issue and lost my backups but thankfully it was only my home lab.  When you upgrade to 2025 the ReFS format changes so when going back you are snookered.  Do what Max says if you need the data otherwise format again.

​@regnor Thank you for this information. I did not stumble across that in my Google searches yesterday.

50TB REFS VOLUME ON S2022.   CONNECTED IT TO W11 TO GRAB SOME FILES AND IT AUTOMAGICALLY UPGRADED TO A FORMAT NO LONGER READABLE ON S2022.    SO I INSTALLED VBR ON W11 AND KEPT GOING BUT THE BOX IS NOT STABLE.  LOCKUPS AND MORE LOCKUPS. CURSOR MOVES BUT NOTHING ELSE RESPONDS.  HAVE TO PULL THE PLUG OUT EVERY TIME I WALK BY. CONNECTED THE VOLUME TO A NEW HYPERV GUEST RUNNING W11 AND THIS TIME, THE GD VOLUME STARTED GETTING ENCRYPTED BY BITLOCKER WITHOUT CONSENT.   I CAUGHT IT AT 17% AND IT’S BEEN DECRYPTING FOR SEVERAL DAYS.   THIS W1124H2 HAS BEEN UNSTABLE FOR WEEKS.  I BOUGHT NEW DRIVES AND AM GOING TO COPY THE DATA ACROSS THE WIRE FROM THE W11 REFS VOLUME BACK TO S2022 REFS WHERE IT WAS CONTENT AND WORKED FOREVER WITHOUT ISSUE. 

Hey ​@Mike D15 i take it you were using fast clone to dedupe your backup chains’ synthetic fulls?

that won’t copy across the wire in the same block cloned state and you may find that it consumes much more storage as a result.

If you use Veeam to move the data between backup repositories, Veeam will do the block cloning again.

​@Mike D15 , Did you fix the problem? if so, what did you do to fix it?

​@AndrePulia I was lucky enough to have an extra 50Tb of space on a Compellent storage that I connected to a S2025 VM as NTFS. I also connected the ReFS drive in question, which was also a LUN on the Compellent,  to that S2025 VM. From there the OS could see the original backup set so I copied it from the ReFS drive to the NTFS drive using RoboCopy. It took nearly 2 weeks to transfer all of the data and get it connected back to the original physical server running S2019. I was able to salvage almost 90% of the backups. The other 10% I didn’t have space for so I opted to ignore a few not important backups sets. There were some hoops to jump thru in Veeam like rescanning repositories and manually reconnecting backup sets to jobs but in the end it is all working again.

I have a new physical server and Dell DataDomain being delivered next week that this will all migrate to. This should not be an issue moving forward and I will not overlook the hardware’s compatibility with new OS’s next time. I learned my lesson the hard way.

Wow quite the fix but glad you got around it to move forward.  Lesson learned as they say. 😁