Describe the bug
When writing a cube with data_store.write_data(ds, output_id, replace=True) by using the s3 data store, the final cube is corrupt.
To Reproduce
Steps to reproduce the behavior:
use xcube datastore for s3 to write a cube into s3 bucket
modify cube, and write it again to the same location by using parameter "replace=True"
view cube, it is corrupt - contains no metadata files
Expected behavior
I would expect, that xcube first removes all files, and then writes the new cube into the same location as the previous one. Making sure no files from the old one corrupt the new cube as well as making sure the metadata exists.
Describe the bug When writing a cube with data_store.write_data(ds, output_id, replace=True) by using the s3 data store, the final cube is corrupt.
To Reproduce Steps to reproduce the behavior:
Expected behavior I would expect, that xcube first removes all files, and then writes the new cube into the same location as the previous one. Making sure no files from the old one corrupt the new cube as well as making sure the metadata exists.