I'm performing a full backup of a directory with duplicity 0.7.17. Slow I/O is the bottleneck (less than 20 MiB/s read) and the directory is large (~5 TiB).

The full backup was interrupted before it could finish. I've restarted it with the same command and duplicity correctly resumed it from the volume where it had been left, however it's re-reading the entire directory from the beginning. This means I'm not saving any time compared to doing the backup from scratch.

How can I instruct duplicity to only read the "new" directories to save on I/O?

Extra information:

  • It's not just reading file metadata: in atop I see disk activity which can only be explained by duplicity reading the actual file content.
  • I'm on fedora 27. The directory is on an ext4 LVM partition spanning 2 virtual SCSI disks on vmware (vmx-10).

1 Answer 1


Yes, it is working that way. It's not ideal, but functional. Please submit a bug report in Launchpad to get this corrected.

From restart_position_iterator in duplicity:

Fake writing to backend, but do go through all the source paths.
Stop when we have processed the last file and block from the
last backup.  Normal backup will proceed at the start of the
next volume in the set.
  • Do we know that status of this now?
    – Diagon
    Mar 29, 2020 at 6:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .