crop VS cloudplow

This is pretty much how copy-on-write ropes work. Check out xi-rope, Ropey or crop, they're all built using B-trees and implement the behavior you described.

It already has that

I might just stick to my previous idea, but it's not user friendly for other people in my household. I'm using Cloudplow. So I'll add an rclone_extras tag for '--exclude-from' that references an excluded list.

Multiple in parallel. To avoid this, you could write a simple bash script to only allow one instance of itself. Example here. Alternatively, I recommend checking out cloudplow.

Something else of interest might be Cloudplow or Crop. Cloudplow is a Python implementation of rclone that extends it to a serviceable state for automatic transfer of files. Crop does this similarly but uses the go language.

Check out Cloudplow and rclone.

You can use something like cloudplow to get around the 750gb limit. Cloudplow can automate using different service accounts which each have their own limit.

As you’ve mentioned you’d like to use rclone, I’d suggest you read up on unionfs or mergerfs and configuring something like CloudPlow or Crop to manage uploads to the cloud for you.

As you have a linux seedbox, I would look into cloudplow.

One way to accomplish this, at least in function, would be to not use the mount to upload. Use something like mergerfs to combine the mount folder and a local folder. Stuff gets written to to the local folder while stuff gets read from both. Then on a schedule use a script to automatically upload stuff. cloudplow is also an option instead of just calling rclone directly to upload. Instead of operating on a fixed schedule it watches the local folder's size and uploads when needed. This helps you not run out of space.