Storage replication orchestration with operator control
DonkeyFleet observes volume-level SnapMirror protection across customer-managed ONTAP and Amazon FSx for ONTAP. It compares declared protection intent with current storage state, explains the drift, and—only when policy and operator authorization allow it—changes the destination.
Its primary use case is bare-metal Kubernetes infrastructure where a CSI driver such as Trident provisions persistent volumes on customer-managed ONTAP. DonkeyFleet protects the underlying ONTAP volumes and their destination copies; it does not replace the Kubernetes PV/PVC lifecycle.
DonkeyFleet is deliberately narrow. It is a controller for a defined fleet of source volumes and destination relationships, not a general storage-management platform and not a failover system.
What it does
- Discovers the ONTAP volumes that provide storage for Kubernetes persistent volumes.
- Registers source and destination clusters and SVM pairs.
- Discovers volumes, SnapMirror relationships, aggregates, schedules, and policies.
- Selects volumes through ordered policy rules and explicit ignore rules.
- Proposes new protection, resize, tiering, orphan-cleanup, and fill-growth work.
- Supports manual, approval-gated, and bounded automatic operation.
- Persists intent, job identity, approvals, notifications, and audit history.
The control path
An approval never calls ONTAP from the browser. It authorizes a plan item. A later reconcile observes the fleet again and executes only if identity, policy, safety limits, and effective dry-run still permit it.
Start here
- Review supported topologies and the safety model.
- Choose the standalone or Kubernetes deployment.
- Keep dry-run enabled and complete the first dry-run.
- Review current limitations before enabling Apply.
Breaking, resynchronizing, or reversing SnapMirror relationships for disaster-recovery failover is permanently outside DonkeyFleet's scope.