DNS drift records a difference between expected management state and observed provider state. Investigate what changed and which system is responsible before requesting reconciliation. An external edit may be intentional, and a remote zone can belong to a hosting account that must not be overwritten.
| Method | Resource | Scope | Idempotency-Key |
|---|---|---|---|
GET |
/v2/dns/zones/{zoneId}/drift |
dns:read |
Not required by this route |
POST |
/v2/dns/zones/{zoneId}/reconcile |
dns:write |
Required |
01 Build the incident picture
- Read the owned zone and page through its dedicated
driftcollection. The embedded list in zone detail is limited and is not a complete history. - Check each event's state and observation time. The dedicated collection can include resolved as well as open events.
- Compare the affected name, type, value and provider binding with the current management records and the provider's actual answer.
- Review recent changesets, imports, restores, hosting provisioning and direct provider edits. Identify the intended final records before choosing a recovery action.
- Check ownership warnings and management locks. A shared provider account does not make every zone available to every tenant.
02 Request a fresh reconciliation
After resolving the cause or approving the intended state, POST to the zone's reconcile endpoint with an idempotency key. Preserve the returned job reference and inspect its outcome. The request schedules work; it is not proof that every server now matches.
Read the drift collection again and compare new timestamps with the earlier evidence. Verify important names directly against the applicable authoritative servers. If reconciliation reports an ownership conflict, fix the legitimate assignment through the administrator's supported process rather than deleting the remote zone to suppress the warning.
Common questions
Does reconciliation always overwrite the provider?
Do not treat it as blanket permission to do so. Provider capabilities, protected records, ownership and the specific reconciliation workflow determine the result. Review the intended state and returned evidence.
Why does an old drift event remain in the list?
The endpoint provides history across states. A retained resolved event documents an earlier difference; compare status and time before reporting it as a new incident.
Can public DNS differ without a current drift event?
Yes. Public delegation may point to another service, a replica may be delayed or a resolver may retain an old answer. Distinguish management records, provider readback, authoritative answers and cached resolver answers when diagnosing the discrepancy.