ContinueFailedRestore

From RdiffBackupWiki

When a restore fails for some reason (ex. network outage), the destination contains a partial copy of the backup. Some files and directories are missing, some files and directories are only partially copied, but some files and directories are fully copied and need no modification.

The current implementation first refuses to restore, then after giving the --force option, overwrites everything, basically doing the whole restore procedure again, copying everything again. This means that if you restore a big directory structure (think about restoring / of a machine, several hundred GB of data) if the restore fails at 90% you have to start from 0%.

A better implementation should traverse the directory tree just like the current, but when finding a file that exists on the destination, syncs it using the clever methods used when doing backup, instead of copying the file again. In the optimal case, the hash of the destination file matches the backup and no copying is needed.

-- AndrasGBekes