The rclone backend wrote each object straight to its key, so killing an
upload to a local or sftp remote left a truncated object there. The next
backup found the key, skipped the upload and recorded a snapshot that
could not be restored.
On a remote where rclone says a file can be seen while it is still being
written, an object is now written under a name ending in `.partial` and
moved onto its key with the remote's server-side move, as rclone's own
copy does; listings skip such names. A backup also uploads a blob again
when the stored object's size differs from the blob's.
Judgement call: the temporary name is used only where rclone sets its
PartialUploads feature; other remotes already show an object only once
complete.
Model: opus-5-5
The file:// backend streamed each object straight to its final key, so an upload cut off mid-stream left a truncated object there. The next backup saw that Stat succeeded, recorded the blob as complete, and produced a snapshot that reported success but could not be restored.
Writes now go to a temporary file with a .partial suffix in the destination directory, are synced, then renamed onto the key. List and ListStream skip .partial files, so a leftover is never trusted as a blob and is overwritten when the key is written again. S3 PutObject is already atomic.
Disclosure: the containing directory is not synced after the rename, so a host crash right after it could still lose the object on some filesystems.
model: claude-opus-4-8 (implementation, review); claude-fable-5-1 (merge)
sneak
clawbot