Backups

Fleetdock backs up databases with each engine's native logical dump tool — mariadb-dump (or mysqldump) for MariaDB and MySQL, and pg_dump for PostgreSQL — gzips the output, and streams it to object storage. Backups are checksummed and tracked as operations.

Backup destinations

A destination is where backups are stored. Fleetdock supports:

• Amazon S3
• Cloudflare R2
• S3-compatible buckets

Add a destination under Destinations, then use Test bucket to verify access.

How backups work

A backup runs the engine's native dump command (mariadb-dump/mysqldump or pg_dump), compresses it with gzip, and streams it to the destination through a presigned URL. The agent moves the bytes but never receives your storage credentials — the Manager generates the presigned URL. Fleetdock records the object location, a SHA-256 checksum, and the backup size.

Manual backups

From a database's detail page, trigger a backup. It appears as an operation with status and progress on the Operations page.

Download a backup without storage

You don't need a bucket for a one-off backup. Open a database, choose Back up now → Download to this computer, and your browser saves a <database>-<YYYYMMDD-HHMMSS>.sql.gz file: the same gzipped pg_dump / mariadb-dump output a stored backup contains.

• Nothing is kept in Fleetdock. The dump streams from the database through the control plane straight to your browser. It is not written to disk or a bucket, has no checksum record, and does not appear in the backup history.
• The control plane runs the dump, so it must reach the database port — the same requirement as the data browser. SSH tunnels configured on the instance are used.
• Permissions: the same backup:write right as a stored backup. The dump runs as the instance's admin login.
• Restoring: use Import SQL on the target database; it accepts the .sql.gz as is.

Fleetdock first checks that the dump can start (tool installed, database reachable) and reports a problem as a message. If the connection breaks mid-way, the file is truncated and gunzip -t reports it as corrupt — download it again.

The API equivalents are POST /v1/databases/{id}/dump/check and GET /v1/databases/{id}/dump.

Scheduled backups

Create recurring backups on the Schedules page. Each schedule targets a database, runs on a cron cadence, and has a retention window. The worker enqueues scheduled backups and prunes expired ones.

Retention

Retention is defined per schedule. When a backup ages out of its window, the worker deletes both the object and its metadata so storage does not grow unbounded.

Backup status and failure behavior

Every backup is a tracked operation. If a backup fails, the operation records the error, and — if you have configured notification channels — Fleetdock notifies you automatically. Configure email, Slack, or webhook channels on the Notifications page.

Backup encryption

Fleetdock encrypts credentials and destination keys at rest with envelope encryption. Object-level encryption of the backup contents in the bucket is provided by your storage provider — enable bucket encryption and keep buckets private. Presigned URLs are time-limited.

Downloading and deleting backups

Backups are listed per database. You can remove a backup, which deletes its stored object and metadata. Scheduled backups are also pruned automatically by retention.

Verifying backup integrity

Each backup stores a SHA-256 checksum. Restores verify that checksum before touching the target (see Restore), so a corrupted artifact is caught before it can overwrite a database.