First dry-run
Dry-run is enabled by default. Keep it enabled until identity, policy, capacity, and proposed work all match your expectations.
1. Register infrastructure
As an administrator:
- Store ONTAP credentials in Vault.
- Register each cluster with its source or destination role.
- Confirm the probe records the physical cluster UUID and ONTAP version.
- Register and activate the intended SVM pair.
2. Record human choices
Create a replication profile with an existing destination aggregate, SnapMirror policy, schedule, and supported destination naming mode. DonkeyFleet discovers these values but does not choose or create them.
Create an automation profile, a volume group, and ordered match rules. Add explicit ignore rules for volumes that must never be managed.
3. Review the run
Wait for the scheduled reconcile, or make a policy change to request an immediate run, then check:
- Home for last-run outcome and warnings.
- Inventory for candidate, adopted, and suspended relationships.
- Approvals for proposed actions and skip reasons.
- Capacity for protected, unprotected, and orphaned totals.
- Notifications and Audit for transitions.
Any incomplete page of ONTAP data must stop evaluation with skipped_observe_incomplete.
4. Establish the trust boundary
Verify the safety model, resolve any skip reasons (ambiguous aggregate, name collision, missing schedule), and review the current limitations. Only an administrator should then decide whether to unlock and disable dry-run.
Approval can be tested while dry-run remains on. Approved work stays visible but will not mutate ONTAP until effective dry-run is false and a later reconcile still considers it eligible.
If onboarding existing volumes, review the automation profile's minimum volume age. Use 0 only
when those volumes should be eligible immediately; otherwise Backlog will remain unchanged until
the age is reached.