BackupSchedule CRD

The BackupSchedule Custom Resource Definition defines cron-based scheduled backups. Each run creates a Backup CR; retention is controlled by history limits.

Overview

API Group: retour.thebackup.company/v1
Kind: BackupSchedule
Scope: Namespaced
Cron format: Standard 5-field cron (minute, hour, day-of-month, month, day-of-week).

Spec

Required Fields

Field	Type	Description
`schedule`	string	Cron expression (5 fields). Example: `0 2 * * *` (daily at 02:00).
`storageBackend`	string	Name of the StorageBackend for backups.

Optional Fields (with defaults)

Field	Type	Default	Description
`namespacedResources`	array	`[]`	Resources to backup (same as Backup).
`clusterResources`	array	`[]`	Cluster-scoped resources.
`excludePVs`	bool	`false`	Skip PV data.
`resourceFilter`	object	`nil`	Include/exclude filters.
`suspend`	bool	`false`	If `true`, pause subsequent runs.
`successfulBackupsHistoryLimit`	int32	`3`	Number of successful backups to retain.
`failedBackupsHistoryLimit`	int32	`1`	Number of failed backups to retain.
`concurrencyPolicy`	string	`Allow`	How to handle overlapping runs.

ConcurrencyPolicy

Value	Behavior
`Allow`	Create new backup even if previous is still running.
`Forbid`	Skip the run if a backup is already in progress.
`Replace`	Cancel the running backup and start a new one.

Retention

successfulBackupsHistoryLimit keeps a flat count. retention replaces it with grandfather-father-son tiers, keeping the newest backup per period up to each count:

Field	Type	Description
`retention.keepHourly`	int32	Newest backup of each hour to keep.
`retention.keepDaily`	int32	Newest backup of each day.
`retention.keepWeekly`	int32	Newest backup of each week.
`retention.keepMonthly`	int32	Newest backup of each month.
`retention.keepYearly`	int32	Newest backup of each year.
`backupDeletionPolicy`	string	Whether expiring a backup also purges its data from the backend. Object Lock retention still defers the purge.

Backups are not owned by their schedule, so deleting the schedule leaves its backups and their data in place.

Restore Drills

drill restores this schedule's newest successful backup into scratch retour-drill-* namespaces on its own cron, optionally runs a check pod against the result, and deletes the scratch namespaces afterwards.

Field	Type	Description
`drill.schedule`	string	Cron expression for drill runs. Independent of the backup cron.
`drill.check.image`	string	Image the check container runs.
`drill.check.command`	[]string	Command whose non-zero exit fails the drill. Restored claims are mounted at `/data/<claimName>`.
`drill.check.namespace`	string	Original namespace whose scratch clone the check runs in. Defaults to the first restored namespace.
`drill.check.timeoutSeconds`	int32	Bounds the check run. Default 300.

The outcome lands in status.lastDrill, a Kubernetes event, and the retour_restore_drill_* metrics. A drill with no volumes to restore is recorded as Skipped, not Failed.

Timing

Field	Type	Default	Description
`startingDeadlineSeconds`	int64	—	How long after a missed fire time a run may still start, with CronJob semantics. Without it, a rollout or API blip that crosses the fire time drops the run. A skipped run is an event and a counter either way.
`suspend`	bool	`false`	Pause subsequent runs.

Inherited by Every Backup

These are copied onto each Backup the schedule creates and behave exactly as they do on Backup: scope, excludedNamespaces, includedSystemNamespaces, includeWebhookConfigurations, resourceDiscovery, namespaceSelector, hooks, databases, secrets, volumeGroups, skipVerification and bandwidthLimit.

Status

Field	Type	Description
`phase`	string	`Pending`, `Active`, `Suspended`, or `Failed`.
`lastScheduleTime`	string	When the last backup was created.
`lastBackupName`	string	Name of the last Backup CR created.
`lastBackupStatus`	string	Status of the last backup.
`nextScheduleTime`	string	When the next backup will run.
`activeBackups`	[]string	Names of backups currently running.
`error`	string	Error message if schedule failed.
`lastDrill`	object	Outcome of the last restore drill: phase, what was restored, and the check result.
`lastDrillTime`	string	When the last drill ran.
`lastSelectedNamespaces`	[]string	Namespaces the last run matched, for schedules that select by label.
`conditions`	array	Standard Kubernetes conditions, including `QuotaExceeded` when a BackupQuota is holding this schedule back.

Examples

Daily Backup at 02:00

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: daily-backup
spec:
  schedule: "0 2 * * *"
  storageBackend: s3-production
  namespacedResources:
    - resourceType: Deployment
      namespace: production
      name: webapp
  successfulBackupsHistoryLimit: 7

Hourly with Forbid (No Overlap)

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: hourly-backup
spec:
  schedule: "0 * * * *"
  storageBackend: s3-production
  concurrencyPolicy: Forbid
  namespacedResources:
    - resourceType: StatefulSet
      namespace: production
      name: database
  successfulBackupsHistoryLimit: 24
  failedBackupsHistoryLimit: 3

Weekly Full Namespace Backup

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: weekly-namespace
spec:
  schedule: "0 3 * * 0"
  storageBackend: s3-production
  clusterResources:
    - resourceType: Namespace
      name: production
  successfulBackupsHistoryLimit: 4

Suspended Schedule

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: paused-schedule
spec:
  schedule: "0 2 * * *"
  storageBackend: s3-production
  suspend: true
  namespacedResources:
    - resourceType: Deployment
      namespace: production
      name: webapp

Manifests Only (No PV Data)

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: manifests-schedule
spec:
  schedule: "0 */6 * * *"
  storageBackend: s3-production
  excludePVs: true
  clusterResources:
    - resourceType: Namespace
      name: production
  successfulBackupsHistoryLimit: 28

GFS Retention with a Weekly Drill

apiVersion: retour.thebackup.company/v1
kind: BackupSchedule
metadata:
  name: nightly
spec:
  schedule: "0 3 * * *"
  storageBackend: s3-production
  startingDeadlineSeconds: 3600
  clusterResources:
    - resourceType: Namespace
      name: production
  retention:
    keepDaily: 7
    keepWeekly: 4
    keepMonthly: 12
  drill:
    schedule: "0 5 * * 0"
    check:
      image: postgres:17
      command: ["sh", "-c", "test -s /data/db-data/PG_VERSION"]