OpenEverest adds time-based retention for scheduled backups

•
openeverest backups releases

“Keep the last 7 backups” and “keep 30 days of backups” are different policies, and OpenEverest v2.0.0-dev.4 supports both. A schedule’s retention is either {type: count, count: 7} or {type: time, duration: 30d}, with the duration in days, weeks or months (Nd, Nw, Nm). Each provider maps it onto its backup engine. Anything the engine can’t do is reported on the instance’s BackupConfigured condition instead of being silently ignored.

Support in the providers released alongside dev.4:

  • PostgreSQL (v0.4.0): count and time, enforced by pgBackRest. Time windows are converted to days, and a month counts as 31 days so you never keep less than you asked for. Differential schedules support count only.
  • MySQL (v0.3.0): count only. The Percona XtraDB Cluster operator prunes old backups and deletes them from storage.
  • TiDB (v0.1.5): count and time, run by the provider’s own backup scheduler.
  • MongoDB and Cassandra: count only. Cassandra applies retention to the whole cluster rather than per schedule.

retention replaces retentionCopies. Once the new CRDs are installed, the old field is dropped and those schedules keep every backup, so re-apply them with retention: {type: count, count: N}. The UI still edits count retention only. See the dev.4 release notes for the full list of changes.

Read the docs
Back to What's New