# Make your next restore a familiar one.

> Prepare an isolated restore destination, choose an encrypted snapshot and validate your recovered database and files. Understand what each backup check proves.

Canonical page: https://www.calmbackup.com/recovery

Select a snapshot, recover it with your key, and check the application. A practical recovery routine starts before an incident.

[Start free trial](https://app.calmbackup.com/register)

## Know what each green light means.

A backup moving successfully through one stage is useful evidence, but it does not answer every recovery question. Keep upload status, archive integrity and application recovery separate.

### Dump check

The driver checks the generated database dump. PostgreSQL lists the archive, MySQL checks for a completion marker, and SQLite runs an integrity check.

### Authenticated decryption

AES-GCM verifies the authentication tag when the client opens an encrypted archive. An invalid key or modified ciphertext causes decryption to fail.

### Application restore test

You restore into a separate environment and check that your application can use the recovered database and files. Include this in your operating routine.

**About automated restore testing**

The CLI workflow described here does not schedule a daily sandbox restore of your application. A confirmed upload or dump check should not be represented as that test.

## Prepare the destination first.

A restore writes to the database and directory paths in the configuration you use. Work on a separate recovery machine or isolated environment for a practice run. Review the target host, database name, SQLite path and directory paths before running the command.

Use the encryption key that matches the selected archive. Confirm that the destination has the database tools and versions needed to import the dump, enough disk space for the downloaded and extracted files, and the application configuration needed for a useful test.

**A restore can replace existing data**

PostgreSQL restores use clean/if-exists behavior; MySQL imports the SQL dump; SQLite copies the recovered database file to its configured path. Included directories are copied back to configured paths. Do not run a practice restore against your production configuration.

## Choose a snapshot. Follow the restore.

List the available snapshots and choose an explicit backup ID. Using an ID makes it clear which point in your backup history is being recovered. Use an explicit configuration path for a dedicated recovery environment.

1.  ### Find the archive
    
    The client retrieves the backup metadata and download URL. If a local encrypted copy matches the recorded checksum, it can be reused; otherwise the archive is downloaded.
    
2.  ### Decrypt and unpack
    
    The client opens the archive with your configured key, checks the GCM authentication tag and extracts the compressed contents locally.
    
3.  ### Restore the data
    
    The database driver imports or copies its dump, then the configured directories are restored to their destination paths.
    
4.  ### Check the application
    
    Confirm representative records, uploaded files, permissions and application startup. Perform a few real reads and writes in the test environment.
    

```
calmbackup --config /path/to/recovery.yaml list
calmbackup --config /path/to/recovery.yaml restore <backup-id>
```

[Read the command reference](https://www.calmbackup.com/docs/commands)

## Make the runbook useful to someone else.

Record where to find the recovery key, how to obtain a backup ID, which configuration to use, and how to validate the result. Keep that runbook accessible if the original server is offline.

Measure the time from starting the download to a working application. Your backup frequency affects how much data you might lose; your restore procedure and environment affect how long you may be unavailable. Neither is guaranteed by a successful upload.

### After a major change

Repeat a restore exercise after a database upgrade, configuration change or new storage path.

### Before a migration

Keep the original backup history and encryption keys until you have confirmed recovery in the new environment.

## Good questions.  
Clear answers.

Can I restore without a local copy?

Yes, the client downloads the selected encrypted archive when a suitable cached copy is not available. You still need access to the backup source and the correct encryption key.

Does this provide continuous point-in-time recovery?

The workflow described here restores a selected backup snapshot. It does not describe transaction-log replay to an arbitrary moment between snapshots.

Should I always restore the newest backup?

Choose the snapshot that matches the incident. A recent backup can already include an accidental deletion or bad application change. Review timestamps and keep enough history for your recovery needs.
