Skip to main content
Database backups are configured from a database’s Backup section. Paid database plans can create daily encrypted recovery points and manual backups. The retention period comes from the plan.

​
Connection

Choose one destination: For a custom S3 destination, enter: Pxxl checks the bucket with the supplied credentials before saving the destination. Access keys and secrets are stored in protected credential storage, not in the database as plaintext.

​
Backup flow

Changing the destination creates a new destination record. A completed backup keeps the destination and credentials that were used when its object was written, so changing settings does not move older recovery points.

​
Configure backups

1

Open the database

Select a database, then open Backup.
2

Choose the destination

Select Pxxl-managed storage or custom S3-compatible storage.
3

Enter and verify the destination

For custom S3, enter the endpoint, bucket, region, prefix, access key, and secret key. Pxxl verifies that the bucket can be reached.
4

Enable daily backups

Turn on daily backups and save. You can also create a recovery point immediately from Recovery points.

​
AWS CLI preflight

Use the AWS CLI to confirm that a restricted key can see the bucket and, if you choose, write a test object:
The last two commands write to and delete from the bucket. Use a test prefix and run them only when you intend to make that change. For AWS S3, use the bucket’s region. For Cloudflare R2, use the account endpoint and AWS_REGION=auto. For MinIO, use its public HTTPS endpoint and configured region.

​
Required bucket permissions

Pxxl needs to verify the bucket, write new recovery points, read them for download and restore, and delete objects when a recovery point is removed or expires.
If you only need to create backups and never download, restore, or remove them from Pxxl, your policy can be narrower. The full set supports the actions available in the Backup section.

​
Recovery points

Each recovery point moves through a small lifecycle: Only completed recovery points can be restored. A restore is queued separately and can replace the current database data. Confirm the selected recovery point before starting it and wait for the restore status before reconnecting the application.

​
Monitoring and troubleshooting

When a backup fails, check the recovery point error, the destination endpoint, the bucket region, and the access key permissions. A common configuration issue is a key that can list a bucket but cannot write to the selected prefix. If the database is healthy but a restore cannot start, check whether another backup or restore is already running. Pxxl allows one active backup and one active restore per database at a time. For the database connection itself, return to Managed Databases. For general S3-compatible setup, see Storage.