# Restore From Backup - Keep Database Each Time

**URL:** <https://forum.duplicati.com/t/restore-from-backup-keep-database-each-time/22281>\
**Category:** Uncategorized\
**Created:** [April 11, 2026, 7:46pm UTC](https://forum.duplicati.com/t/restore-from-backup-keep-database-each-time/22281 "2026-04-11T19:46:05Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![sgeklor](https://forum.duplicati.com/letter_avatar_proxy/v4/letter/s/eada6e/32.png) [@sgeklor](https://forum.duplicati.com/u/sgeklor)\
**Post date:** [April 11, 2026, 7:46pm UTC](https://forum.duplicati.com/t/restore-from-backup-keep-database-each-time/22281/1 "2026-04-11T19:46:06Z")

</div>

I have a 4.6 TB backup collection that I wish to restore.

Because the restore operation is taking a long time, there have been multiple interruptions along the way. I am aware that the restore can be restarted by choosing the same output folder and choosing “overwrite existing files” rather than versioning files.

However, each time I restart the restore the database has to be recreated and this takes a long time to complete.

Is there any way the database can be cached? Perhaps save it to tmp folder with a filename that is derived from the backup source in some way?

---

<div class="post-metadata">

**Author:** ![ts678](https://forum.duplicati.com/letter_avatar_proxy/v4/letter/t/8491ac/32.png) [@ts678](https://forum.duplicati.com/u/ts678)\
**Post date:** [April 13, 2026, 11:19pm UTC](https://forum.duplicati.com/t/restore-from-backup-keep-database-each-time/22281/2 "2026-04-13T23:19:55Z")

</div>

> [@sgeklor](#):
>
> Is there any way the database can be cached?

I don’t think feature exists. As UI says (or maybe used to), it’s a temporary file.

> [@sgeklor](#):
>
> Perhaps save it to tmp folder with a filename that is derived from the backup source in some way?

You can make a dummy job with basically any Source config, just to do restore.  
Be careful not to do a backup or any maintenance that might affect Destination.

> [@sgeklor](#):
>
> I am aware that the restore can be restarted by choosing the same output folder and choosing “overwrite existing files” rather than versioning files.

I think the first part is all you need. I don’t think it overwrites existing correct files.

If you originally like “Save different versions with timestamp in file name”, keep it.

Rationale is below, and if I misread new restore flow, maybe devs will correct me.

[Restore Only Differences of Files Between Two Versions?](https://forum.duplicati.com/t/restore-only-differences-of-files-between-two-versions/22231/7)

(see the part about checking the target files first to see if any work must be done)

EDIT 1:

You just recreate the database of the dummy job from a destination, then restore.

EDIT 2:

DB recreate will be slower, especially if you have a lot of versions. A workaround limiting to 1 version is CLI with `--version` option, but that’s more work to set up.