Run a Velero backup
Take a point-in-time backup of an application (its namespace, resources, and volume data) on an edge cluster — and restore from it.
Prerequisites
- Velero installed on the target edge — the standard edge preset includes it
- Somewhere for the data to go: the cluster's own
BackupStorageLocation, and/or an external backup location defined for your org - Permission to update edge instances in your role
Take an ad-hoc backup
Sidebar → Applications → click the release → Backups tab → Take Backup.
The dialog has two fields:
- Retention (TTL) — how long Velero keeps the backup before
auto-deleting. Default
720h(30 days); any Go-style duration works (24h,30m) - Backup Location — only shown when your org has at least one backup location defined. The default option is your org's default location if one is marked, otherwise In-cluster storage (default). External locations survive the loss of the cluster — pick one for backups you'd want after a cluster disaster.
Everything else is automatic: the backup is scoped to the
release (its namespace + its resources, volume data included when
the cluster has CSI snapshot support), and the name is generated
as <release>-backup-<timestamp>.
Submit. The new row appears with status New → InProgress →
Completed. Manifest-only apps complete in seconds;
volume-carrying apps (databases, VMs) take as long as the data
transfer needs. While it runs, the tab's action buttons are
disabled — one Velero operation per release at a time.
Schedule recurring backups
Same Backups tab → Backup Schedules → Create Schedule.
- Cron Expression — standard 5-field cron
(e.g.
0 2 * * *for 2am daily) - Retention (TTL) — every run inherits this; older runs auto-delete when it expires
- Backup Location — same dropdown as the ad-hoc dialog, when external locations exist
Velero owns the cron on the edge; each scheduled run shows up as a row in the backups list. To stop a schedule, delete it from the drawer — backups it already produced stay until their own TTL.
Restore from a backup
Backups list → Restore on a Completed backup → confirm.
Restore always targets the release's own namespace and brings the app back to the state captured in the backup — resources and volume data alike. It runs as an async operation; watch it in the list or the Operations Center.
Where backups land
Two possibilities, depending on what you picked in the Backup Location dropdown (or its default when you left it alone):
- In-cluster storage — Velero writes to the
BackupStorageLocationthe cluster was provisioned with. DT Edge Platform doesn't manage the bucket itself — it just reads and creates Backup / Restore CRs. - An external backup location — an S3 target your org (or the platform admin) defined under Organizations → Backup Locations. These survive even if the cluster itself is lost. See How-to: manage backup locations.
Before pointing important schedules at an external location, use its Test action (Organizations → Backup Locations → row → Test) to confirm the target edge's backup agent can reach it — the verdict comes back as Available / Unavailable and can take up to 90 seconds.
Common errors
- Invalid TTL / cron format — TTL must be a duration like
720h; cron must be 5 fields OPERATION_IN_PROGRESS— another operation (install, upgrade, another backup…) is running on this release; wait for it to finishBackupStorageLocationnot Available — Velero can't reach the storage backend. Check the BSL's status on the cluster:kubectl -n velero get backupstoragelocationBACKUP_LOCATION_UNAVAILABLE— the external backup location you selected is unreachable from this edge; check endpoint, credentials and CA certificate, or run its Test actionBACKUP_LOCATION_NOT_FOUND— the location was deleted after you opened the dialog; pick another one. Backups already stored in a deleted location remain restorable.PartiallyFailed— most resources got captured but at least one item errored; check Velero's logs on the edge
See also
- How-to: manage backup locations
- Reference: Backups tab
- Velero upstream docs: https://velero.io/docs/