---
title: Restore the console
description: "Restore the console from a console backup, on a new host after the loss of the console host or in place, and check the servers and services."
sidebar:
  order: 17
related:
  - /console-backups
  - /operations/console-outage
---

A restore returns the console to the state of a
[console backup](/docs/v2026.11.5-beta/console-backups). Servers keep their mesh identities and
overlay addresses. Database services do not roll back.

| Case | How |
| --- | --- |
| The console host is lost, or the console does not start | [Restore on a new host](#restore-on-a-new-host) with the installer. The same steps restore over a damaged console on the same host. |
| The console runs, and you want an earlier state | [Restore in place](#restore-in-place) in the console |

While the console is down, a server that restarts cannot join the mesh (see
[Console outage](/docs/v2026.11.5-beta/operations/console-outage)). Restore the console soon.

## Be ready before a failure

- Set up [console backups](/docs/v2026.11.5-beta/console-backups) with an S3 target off the
  console host. Keep the recovery key file off the console host too. Without
  it, nobody can open a backup.
- Use a DNS name in `TSE_PUBLIC_URL`. A restore on a new host is then a DNS
  change.
- Keep a standby Linux host with Docker outside the failure domain of your
  services (see [Requirements](/docs/v2026.11.5-beta/installation#requirements)). On an airgapped
  site, keep the airgapped package of the console release on it.

## What you need

- **The backup**: the newest `.tse` file of this console
  (`tse-console-<console>-<time>-<release>.tse`), as a local file. While the
  console runs, select **Download** on it in **Administration → Console
  backups**. Else get it from the backup directory (`TSE_BACKUP_DIR`) or
  from the S3 bucket.
- **The recovery key file** of the backup
  (`tse-recovery-key-<console>-<date>.txt`). After a key rotation, an older
  backup needs the older key file.
- **The installer** of the release of the backup, or of a newer release.

## Restore on a new host

1. **Stop the old console**

    The old console must never run again: two consoles with the same mesh
    keys both claim the servers. Power off the old host, and prevent its
    start or its connection to the servers. Or select **Uninstall** in the
    installer menu on the old host (see [Uninstall](/docs/v2026.11.5-beta/uninstall)).
    Uninstall removes the console volumes and `.env`.

    If you cannot keep the old console isolated, stop here.

    For a restore over a damaged console on the same host, skip this step.

2. **Copy the files to the new host**

    Copy the backup file and the key file to the new host. Make the key file
    readable only by you (`chmod 600`).

3. **Select Restore the console from a backup**

    On the new host, run the installer:

    ```bash
    curl -fsSL https://enterprise.tigerdata.com/install/timescale-enterprise.sh | sh
    ```

    From a bundle, run `./timescale-enterprise.sh` in the bundle directory.
    In the menu, select **Restore the console from a backup**. At
    `Console backup file:`, enter the path of the `.tse` file.

    To restore without the menu, add `--restore FILE` after `sh -s --` (or
    after `./timescale-enterprise.sh`). Add
    `--offline` on a host with no internet access. To use an install
    directory other than `/opt/tse`, add `--dir` (see
    [Install directory](/docs/v2026.11.5-beta/installation#install-directory)).

4. **Give the recovery key file**

    At `Recovery key file:`, enter the path of the key file. With no
    terminal, set `TSE_INSTALL_RESTORE_KEY_FILE` to the path.

5. **Check the summary**

    The **Console backup** box shows the backup time (`Created`), the
    release, the console, the tiers, and the number of servers and services.

    ![The Console backup summary of the installer during a restore](/docs/screenshots/v2026.11.5-beta/restore-the-console/installer-summary.light.webp)

    The installer stops and changes nothing when:

    - the key does not open the backup;
    - a file of the backup is changed or missing;
    - the backup is of a newer release than the installer;
    - the backup is of a demo and you did not add `--demo` (on the host of
      its demo servers), or the reverse.

    Correct the cause and run the installer again.

6. **Keep or change the server-facing URL**

    At `URL [<URL of the backup>]:`, press Enter to keep the URL, or enter a
    new URL. With no terminal, `TSE_INSTALL_PUBLIC_URL` sets a new URL. See
    [The console address](#the-console-address) for what each choice needs.

7. **Let the installer check for the old console**

    If the old console answers at `<URL of the backup>/api/v1/info`, the
    installer stops. Stop the old console first. Add `--force` to skip this
    check only when you are sure that the old console cannot run again.

8. **Wait for the console**

    On a new host, the installer asks for the
    [backup directory](/docs/v2026.11.5-beta/installation#install) (`TSE_BACKUP_DIR`). Over a
    console on this host, it first backs up that console to
    `<TSE_BACKUP_DIR>/pre-restore/`. If that backup fails, it starts that
    console again and stops (for a damaged console, see
    [Console outage](/docs/v2026.11.5-beta/operations/console-outage#recovery), step 5). At the
    end, it shows **After the restore**.

    If the installer stops, run it again. It continues the restore.

9. **Log in**

    Make the changes in [The console address](#the-console-address), then
    log in with the users, single sign-on and `TSE_AUTH_TOKEN` of the backup.
    Complete [After the restore](#after-the-restore) before other changes.

## The console address

Each server finds the console at the host of `TSE_PUBLIC_URL`, on port
`8090`.

| Case | What happens | What you do |
| --- | --- | --- |
| Restore in place, on the same host | Nothing changes. | Nothing. |
| New host, `TSE_PUBLIC_URL` is a DNS name | The servers connect again about one minute after the DNS change. | Point the DNS record to the new host. Open `8090/tcp`, `41641/udp`, `3478/udp` and the console ports to the new host. |
| New host with the IP address of the old host | As for a DNS name. | Move the address to the new host. |
| New host with a new IP address, and `TSE_PUBLIC_URL` is an IP address | The servers cannot reach the console. | Keep the old address on the new host (a second address, or a forward of `8090/tcp`, `41641/udp` and `3478/udp`). Or set a new URL and run **Re-enroll command** on each server. |

For a new URL, follow the server steps in
[Change IP addresses](/docs/v2026.11.5-beta/operations/change-ip-addresses#the-console-gets-a-new-address).
The installer already set the URL: skip the URL change and restart there.

When the host name changes, a certificate in `./certs` (`provided` mode)
must have the new name. A single sign-on provider must accept the new
redirect URL.

## What a restore rolls back

| Change after the backup | After the restore | What you do |
| --- | --- | --- |
| A server enrolled | Unknown to the console and the mesh. Its mesh traffic stops. | Enroll it again. A member of a service waits for [Replace a server](/docs/v2026.11.5-beta/operations/replace-a-server). |
| A server enrolled again with a new mesh key | The mesh has the old key. | Run **Re-enroll command** on the server again. |
| A service was created | It runs, but the console cannot manage it. | Contact Tiger Data support. **Back up after changes** keeps this window short. |
| A service or a server was deleted | The console shows it again. Its operations fail. | Delete it again. |
| A password, a setting or a license changed | The console has the old value. | Make the change again. Install the license file again. |
| A console backup target, schedule or recovery key changed | The console has the old values. | Make the change again. |

After any restore, every user logs in again. An operation that was running
at the time of the backup shows as interrupted. Check the real state of the
service before you run it again.

## After the restore

Admins see the banner **Restored from the backup of** *time*. Select
**Open the checklist** to open the **After the restore** card. The card
lists unhealthy services and missing server packages. **Dismiss** removes
the banner and the card for all admins.

1. **Test each server**

    In the card, select **Test each server**. A server enrolled after the
    backup does not show or answer.

2. **Check each service**

    On the **Overview** of each service, check that it is healthy. If not,
    use the matching runbook.

3. **Load the missing packages**

    In **Administration → Server packages**, load the missing packages with
    **Upload server packages**, or with `--pack` (see
    [Load server packages](/docs/v2026.11.5-beta/installation#load-server-packages)). A backup
    without the **Server packages** tier has none.

4. **Check the console backups**

    In **Administration → Console backups**, check the targets, schedules
    and recovery key. Make a backup and wait for its check to pass.

5. **Make the later changes again**

    Make the changes from after the backup again.

## Restore in place

You must be an admin. The console must have a confirmed
[recovery key](/docs/v2026.11.5-beta/console-backups#the-recovery-key): the restore first backs up
the current state to it.

1. **Select the archive**

    Open **Administration → Console backups**. In the **Backups** card,
    select **Restore** on an archive. Or select **Restore from a file**, and
    upload a `.tse` file of this console.

2. **Paste the recovery key**

    In **Recovery key**, paste the key file or its `AGE-SECRET-KEY-1...`
    line. For an archive on a target, the dialog names the fingerprint of the
    key it needs.

3. **Check the archive**

    Select **Check the archive**. The console does not keep the key.

4. **Read the summary**

    Read the summary and the table of what the restore rolls back. If the
    check fails, do not continue.

    ![The restore panel after Check the archive, with the summary of the backup](/docs/screenshots/v2026.11.5-beta/restore-the-console/in-place-summary.light.webp)

5. **Restore the console**

    Type the name of this console, and select **Restore the console**.

The dialog shows *The console is backing up its current state…*, then *The
console is staging the archive…*, then *The console is restarting. Log in
again when it is back.* The current state goes to
`<TSE_BACKUP_DIR>/pre-restore/` (the newest 3 archives stay). If that backup
fails, the restore stops and nothing changes.

The restart takes one to two minutes. The UI and the API are down. When the
console is back, log in and complete [After the restore](#after-the-restore).
If it does not come back, follow [Console outage](/docs/v2026.11.5-beta/operations/console-outage).

**Restore** shows only on an archive of this console that is on its target
and did not fail its latest check. The console refuses a restore in place:

- while an operation, a backup, an upload or another restore is running;
- of an archive of another console, of a newer release, or of a demo into a
  production console (or the reverse);
- when the archive has another `TSE_DB_ENCRYPTION_KEY` or `TSE_MESH_SUBNET`
  than `.env`. Use [Restore on a new host](#restore-on-a-new-host) on this
  host instead;
- when the archive does not match its sidecar (size or SHA-256). Run
  **Check** on it, or restore another archive.

To undo a restore in place, copy the archive from
`<TSE_BACKUP_DIR>/pre-restore/` to your computer. Restore it with **Restore
from a file** and the recovery key that opens it. Then complete
[After the restore](#after-the-restore).
