---
title: "Restore files and systems"
description: "Restore files from an Outpost backup to their original place, another folder, or another endpoint, download them as a zip, or export a Windows image as a VHD."
url: "https://support.outpostplatform.com/backup/restore/"
product: "depot"
type: "how-to"
status: "draft"
ai_summary: "The restore wizard on an endpoint's Backups tab: pick a snapshot, choose files or image partitions, then restore to the original location, another folder, another endpoint with user mapping, a zip download (2 GB default cap), or a VHD export of a Windows image. Backup > Restores lists every restore. End users restore their own files from the tray."
availability:
  plans: ["home-lab", "essentials", "enterprise", "msp"]
  platforms: ["windows", "linux", "macos"]
  permission: "backup:restore"
  preview: false
source: "https://gitlab.com/outpostplatform/docs/-/edit/main/src/content/docs/backup/restore.md"
license: "CC BY 4.0"
---

# Restore files and systems

The restore wizard puts files from an Outpost backup back where you need them:
in their original place, in another folder, on another endpoint, or in your
browser as a zip. It also exports a Windows system image as a VHD. This page
walks through the wizard, then covers the **Restores** list and the restores
people start themselves from the tray app.

## Before you begin

- You need the `backup:browse` and `backup:restore` permissions. The Owner,
  Support Manager, and Field Technician roles hold both.
- Restores onto an endpoint run on that endpoint. A restore you start while
  it is offline waits, and starts when the endpoint connects.
- If the client uses zero-knowledge keys, you can restore only onto the source
  endpoint, and the file list comes from that endpoint, so it must be online.
  Zip downloads and restores onto another endpoint are not available. See
  [Backup encryption keys and recovery kits](/backup/encryption-keys/).
- **Replace the current file** overwrites files on the endpoint. Restoring a
  whole partition to its original location overwrites that volume.

## Open the restore wizard

1. Open the endpoint and select its **Backups** tab.
2. Select **Restore**. To start from a particular snapshot, select **Restore
   from here** on it in the **Snapshot timeline**.

The wizard has five steps: **Point in time**, **Files and folders**, **Where
to**, **Confirm**, and **Progress**.

## Pick a point in time

1. Under **Pick a point in time**, narrow the list with **Copy** (**Primary**
   or **Secondary**) and **Kind** (**Files** or **System image**) if you need
   to.
2. Select a snapshot. Snapshots are grouped by day, with the copy, kind, size,
   and file count. A snapshot whose last check failed shows **Verification
   failed**.
3. Select **Next**.

## Choose files and folders

1. Under **Choose files and folders**, expand folders with the **+** button and
   tick what to restore. A ticked folder is restored whole, with everything in
   it.
2. To restore an earlier version of one file, select **Compare versions**. The
   **Versions of this file** list shows every snapshot that holds the file,
   newest first. Select **Use this version** on the one you want.
3. Select **Next**.

For a system image, the first level of the tree is the disk's partitions.
Expand an NTFS partition to pick files inside it. To export the whole image as
a VHD, leave everything unticked.

## Choose where it goes

Under **Where should it go?**, choose a **Restore target**:

| Target | What it does |
| --- | --- |
| **Original location** | Puts the files back where they were on this endpoint. |
| **Another folder on this endpoint** | Restores into the folder you enter in **Folder to restore into**, such as `D:\Restored`, and leaves the originals alone. |
| **Another endpoint** | Restores onto a different endpoint in the same client, such as a replacement computer. |
| **Download a zip** | Downloads the files to your browser through Outpost. |
| **Export a VHD** | Writes the whole disk of a Windows system image as a fixed VHD file, to the path you enter in **Where to write the VHD**. |

For the first three targets, choose what happens in **When a file is already
there**:

- **Keep both copies**, the default
- **Replace the current file**
- **Leave the current file**

### Restore onto another endpoint

1. In **Find the endpoint**, enter part of a hostname or user, and select
   **Search**.
2. Select the endpoint to restore onto.
3. Under **User folders**, map each person's folders on the source endpoint to
   an account on the target. Select **Add a mapping**, then fill in **From
   user**, optionally **From SID (optional)**, and **To user**. Without a
   mapping, files go to the same paths they had.

The target endpoint receives the key for this one restore from Outpost, so it
can read the source endpoint's backup.

### Download a zip

The zip is built by Outpost and streamed to your browser. A selection larger
than `backup.browser_download_max_bytes`, 2 GB by default, is refused with
**The selection is ... bytes; downloads from the browser stop at ...
Restore it to an endpoint instead.** Every download is recorded in the
[audit log](/depot/audit-log/).

### Restore from a system image

- **Export a VHD** works with Windows system images only. Enter a file path,
  or a folder, where the VHD is written as `disk<number>.vhd`. The path can be
  local or a network path the endpoint can reach.
- A whole partition restores to **Original location**, which overwrites that
  volume in place, on Windows only. The volume Windows is running from cannot
  be restored in place. Export a VHD instead.
- A whole partition restores to **Another folder on this endpoint** as a raw
  image file, `disk<number>-partition<number>.img`.
- Files inside an NTFS partition restore like files from a file backup.

## Confirm and start

1. Under **Confirm**, check **From**, **What**, and **To**.
2. Select **Start restore**, or **Download** for a zip.

**Restore progress** shows the state, a progress bar, and the files and bytes
restored. **Waiting for the endpoint** means the endpoint is offline. To stop a
running restore, select **Cancel restore**, then **Stop restore**. Files
already written stay where they are.

## Verify

- The restore shows **Succeeded** under **Restores** on the endpoint's
  **Backups** tab and on **Backup** > **Restores**. **Partial** means some
  files could not be restored, and the progress line says how many.
- The files are on the target endpoint, in the place you chose.

## Track every restore

**Backup** > **Restores** lists every restore on the client,
with **Asked for**, **From endpoint**, **To**, **Items**, **State**, and
**Progress**. Filter by **State** to find running or failed restores, and
select **Cancel** on a restore that has not finished.

Restores people start from the tray app are listed too, marked
**Self-service by** and the person's name. They cannot be canceled from the
console.

## Restores from the tray app

When `backup.self_service_restore` is on, which is the default, people restore
their own files from the [tray app](/tray/): **Backup** > **Restore Files…**.
They see only their own folders. Restored files go to their original folder or
to a **Restored files** folder, and never replace a file that is already
there: a clash is saved beside it as `name (restored).ext`. No file names from
these restores are sent to the console. See
[Use the tray app](/tray/using-the-tray/) for the steps an end user follows.

To turn self-service restore off, set `backup.self_service_restore` to off in
a profile. If any profile at the same rank turns it off, it is off.

## If something goes wrong

| Symptom | Cause | Fix |
| --- | --- | --- |
| **Restore** is unavailable with **Restoring needs the backup:browse and backup:restore permissions.** | Your role lacks one of them. | Ask an owner for the permission. |
| **Another endpoint** or **Download a zip** cannot be chosen, with a message about zero-knowledge backup keys | The client uses zero-knowledge keys. | Restore onto the source endpoint, or use the recovery kit. |
| The file list says the backup is on a network share only the endpoint can reach | The destination is a Windows network share. | Bring the source endpoint online. It lists its own backup. |
| A restore stays at **Waiting for the endpoint** | The target endpoint is offline. | Start the endpoint. The restore begins when it connects. |

## Next steps

- [Monitor backup coverage](/backup/coverage/)
- [Backup encryption keys and recovery kits](/backup/encryption-keys/)