> ## Documentation Index
> Fetch the complete documentation index at: https://docs.axottle.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Backups and restoration

> Agreed backup, separate master key storage, recovery on another server and return of access to the owner.

The backup of Telepúrte should contain a consistent database and an encrypted installation state. To do this, use the regular commands `backup create`, `backup verify` and `backup restore`. Simply copying one file of a running SQLite database does not replace this procedure.

## What to save

| Material | Included in `backup create` | Purpose |
| - | - | - |
| SQLite and `state.enc` | Yes | Users, accesses, assignments, history and identity of |
| Kit Manifesto | Yes | Consistency and Integrity Check |
| `master.key` | No | Decoding data from the original |
| `config.json` | No | Address, path and service settings |
| TLS and reverse proxy | configuration No | Restoration of external input |

Store the master key in a separate, secure way. Without the source key, you can't recover the encrypted backup data, and dashboard recovery also doesn't automatically return the application files on the managed nodes -- they're backed up separately.

## Create and verify a copy

The command can work when the dashboard is running. Prepare a private parent directory available to the `telepurte` user with `0700` mode. The directory of a particular kit does not have to exist yet.

```bash theme={null}
sudo -u telepurte telepurte backup create \
  --config /etc/telepurte/config.json \
  --output /secure/telepurte-backups/2026-10-30

sudo -u telepurte telepurte backup verify \
  --config /etc/telepurte/config.json \
  --source /secure/telepurte-backups/2026-10-30
```

Path and date are an example; for the next copy, use a new name. Verification checks database, status and stored encrypted materials. Save the result with installation ID and time, but don't publish the kit itself.

The success of the `verify` confirms the integrity of this kit. It does not prove that you have chosen the latest backup: separately check the date and the desired recovery point.

## Restoration of installation

1. Stop Telepúrte and other processes that write to its data.
2. Save the current data directory separately, even if the installation is damaged.
3. Prepare the configuration on the target server and the original `master.key` with the correct owner and `0600` mode.
4. Specify an empty private `data_dir` or an as-yet-unexisting directory. The command does not overwrite an existing installation.
5. Check the set and installation ID.
6. Perform recovery with the expected ID.

```bash theme={null}
sudo -u telepurte telepurte backup restore \
  --config /etc/telepurte/config.json \
  --source /secure/telepurte-backups/2026-10-30 \
  --expect-instance tp_REPLACE_WITH_VERIFIED_INSTANCE_ID

sudo -u telepurte telepurte database status --config /etc/telepurte/config.json
sudo -u telepurte telepurte check --config /etc/telepurte/config.json
```

Replace the conditional `tp_REPLACE_WITH_VERIFIED_INSTANCE_ID` with an ID from the tested kit. Do not run the original and restored installation at the same time as the same identity. After successful verification, run the service and check the network, domain, TLS, bastions, agents and integrations.

The old compatible kit can migrate on recovery; the original copy remains unchanged; the base with the newer unknown circuitry should not be opened by the old binary.

## Unfinished recovery

`.restore-incomplete` may remain in the event of a failure. This marker closes the launch of partially recovered data. Do not delete it to bypass the error: stop the service, save the directory for parsing, prepare a new empty directory and repeat the recovery from the tested kit.

If the error is related to the key or installation ID, fix the selected materials first. Creating a new master key will not decipher the old backup.

## Restoration of the owner

If the local owner has lost a password or a second factor, use access to the panel server. there is no HTTP reset of the owner. Stop the service, make an agreed copy and verify the ID through the `telepurte identity` or manifest.

Prepare a new password in the protected file and transfer via stdin from the administrative shell:

```bash theme={null}
sudo -u telepurte telepurte owner recover \
  --login owner \
  --expect-instance tp_REPLACE_WITH_VERIFIED_INSTANCE_ID \
  --config /etc/telepurte/config.json < /secure/new-owner-password.txt
```

Use the original owner's actual login. Restore recalls the original owner's browser and gateway sessions, previous TOTP and backup codes. Reconfigure the MFA when you log in and save the new codes.

Roles, grants, playbooks and schedules are not cancelled by this team. If the cause was compromise, separately review the issued rights and operating decisions before returning the installation to work.

## Verification

Check `database status`, `check`, installation ID and input. Run test SSH, SFTP and reading task. Verify the history, roles and relevance of sources. Recover periodically in an isolated environment without simultaneously connecting the duplicate to working agents.

## Rollback

Until recovery, save the original data and configuration. To return, stop the restored installation and return the agreed old set with the appropriate binary and key. Changes made after the date of the selected copy are absent; they can not be restored with an increase in the shelf life.

The commands refer to a standard Linux installation; the actual paths are checked in your configuration.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.