I have a Spark job where I was trying to rename MapR-FS files across volumes, and it gave me this INFO log:
18/03/07 12:52:10 INFO MapRFileSystem: Cannot rename across volumes, falling back
on copy/delete semantics
A rename would be an atomic operation that produced no extra files in the destination area. Can someone confirm to me that this fallback mechanism is implemented similarly? I.e. if it has any work-space for its files, its external to the target directory?