If the process dies after deleting chunks / the upload node and the second flush during finalization, S3 chunk objects written before the compare-and-set can be orphaned.
The upload bucket lifecycle policies already clean these up eventually, but a dedicated fix (or the planned S3 orphan scanner) would close the gap sooner rather than relying on lifecycle expiry.
If the process dies after deleting chunks / the upload node and the second flush during finalization, S3 chunk objects written before the compare-and-set can be orphaned.
The upload bucket lifecycle policies already clean these up eventually, but a dedicated fix (or the planned S3 orphan scanner) would close the gap sooner rather than relying on lifecycle expiry.