Configure backups

This page is the order to set things up in. To choose a backup type or storage, or to decide whether you need point-in-time recovery, start with Backup and restore.

Configure storage

Required for every backup type. Choose among supported storage types:

• Amazon S3
• Google Cloud Storage
• MinIO and S3-compatible storages
• Microsoft Azure Blob storage
• Alibaba Cloud OSS
• Oracle Cloud Infrastructure Object Storage
• Remote file server

PVC snapshots also need an entry here. They do not upload database files to object storage, but PBM stores backup metadata there.

Enable optional capabilities

Turn these on before the backups you intend to rely on, not after.

• Enable point-in-time recovery — the Operator saves oplog only while PITR is enabled and only once a full backup exists. Enabling it after an incident is too late.
• Enable server-side encryption — encrypts backup data at the storage backend.

Take a backup

• Scheduled backup — recurring protection.
• On-demand backup — a one-off copy before a change.
• Configure PVC snapshots — volume snapshots instead of an upload, for large datasets.

Restore

• On the same cluster
• On a new cluster
• On a new cluster with different replica set names
• Restore a collection under a different name
• Restore to a point in time — requires step 2.
• Restore from a PVC snapshot

Manage retention and deletion

• Configure retention policy — control how many backups to keep with backup.tasks.retention.
• Delete an unneeded backup — deleting the object also removes its files from storage.

Advanced setups

• Multiple storages for backups - more than one bucket.
• Move an external database to Kubernetes - bring data from outside Kubernetes

Next steps

Configure storage

---

Last update: September 2, 2026
Created: September 2, 2026