Sync & restore
Job models
Section titled “Job models”| Model | Direction | Use for |
|---|---|---|
| copy | one-way | seeding archives, exports |
| sync | one-way | workstation -> Drive backups |
| bisync | two-way (enrolled) | working folders kept in sync on purpose |
| mount | foreground | treating Drive as a filesystem |
Sync/account jobs require an enrolled absolute rclone_runtime; the mount path
stays a separate foreground compatibility flow so long-running syncs and
interactive mounts never fight over the same runtime.
Filters
Section titled “Filters”.driveignore (plus config excludes) is the deletion guardrail — author it before
the first sync. Anything not matched is a candidate for transfer and deletion
propagation, so treat filter changes as production changes.
Restore
Section titled “Restore”Restores are pull-based: point better-drive (or rclone directly) at the same remote and read back. Because cleanup operations are quarantine-first and reversible, deleted-by-policy content is recoverable from quarantine rather than lost.
Upgrading
Section titled “Upgrading”Replace the binary and keep the previous one as rollback. Pin upgrades on servers the same way you would any release artifact: verify the checksum, install, read back the version. See the full configuration and command reference.