When issuing a sync and a previous checkpoint exists, nodes manually purged from the cloud drive don't get updated. Workaround: use `sync --full` until fixed.