How do I drop all real files after checking them out from DVC, keeping only the stubs?

31 views Asked by At

DVC, or data version control, allows me to stub out large files with an MD5 hash, push them to a remote store, and version control the hash.

Then, I can checkout those files from the remote and have them locally.


However, sometimes those files are big and my preference when managing many repos with large files is to only have large files checked out in the project I am working on now. Otherwise, I would like to have my local disk storage freed up with the large files only in the cloud, or only on my local machine in compressed form if they must (ideally no: the files may already be compressed, and additional compression will not shrink them).

I have existing scripts that do this manually, although it is complicated to "clean up" correctly in all cases, and I'd like to forget about them.

What is the DVC procedure or command to "clean up" all checked out content or otherwise remotely controlled data s.t. none exists locally (and disk space is freed) but all *.dvc files are retained?

0

There are 0 answers