About backups
OpenEverest gives every technology the same backup workflow — storages, on-demand backups, schedules, point-in-time recovery and restores — while the provider decides how the data is actually backed up, for example with pgBackRest for PostgreSQL or Percona Backup for MongoDB.
note
Backup support depends on the provider. Most database providers support backups, and many support point-in-time recovery. Check the provider's capabilities in the Plugin Hub or in its README.
Building blocks
- Backup storage — a bucket or other place to write backups to, defined once per namespace and shared by the instances in it. Its type is a backup storage class; OpenEverest ships
s3for S3-compatible object storage. See Backup storage. - Backup class — describes how backups are taken for one or more providers. Providers install their backup classes, so you normally just pick one. List them with
kubectl get backupclasses. - Backup — one backup of an instance, taken on demand or by a schedule.
- Schedule — a cron schedule on an instance that writes backups to one storage, with optional retention.
- Point-in-time recovery (PITR) — continuous upload of the instance's transaction log to a storage, so you can restore to any moment within the recovery window.
- Restore — applies a backup or a point in time to an instance.
What you can do
| Task | Where |
|---|---|
| Add a bucket for backups | Settings > Backups storages — Backup storage |
| Take a backup now, or schedule backups | The instance's Backups tab — Create backups |
| Enable point-in-time recovery | The instance's Backups tab — Point-in-time recovery |
| Restore an instance from a backup or a point in time | Actions > Restore from a backup — Restore a backup |
| Create a new instance from a backup | Actions > Create DB from a backup — Create an instance from a backup |
| Create an instance from a backup that is already in a bucket | Import database — Import a backup from storage |
Recommendations
- Store backups outside the Kubernetes cluster that runs the instances, ideally in another region or account.
- Use a schedule rather than relying on on-demand backups, and set retention so the bucket does not grow forever.
- Enable point-in-time recovery for instances where losing even a few hours of data is not acceptable.
- Test your restores regularly by creating a new instance from a backup.