Infrastructure drift control
Explain which system owns each infrastructure field
Drift often returns because two teams or controllers believe they own the same setting. Hand over those boundaries alongside the code and state location.
In this article
Map the controllers
List the infrastructure tool, deployment service, autoscaler and any administrative process that can change managed resources. Identify the fields each is expected to control.
An autoscaler changing desired capacity can be intentional. A deployment repeatedly resetting it may indicate an ownership conflict rather than a broken autoscaler.
Keep exceptions narrow and explained. A broad ignore rule without an owner leaves future maintainers unable to tell expected change from an incident.
Provide the investigation path
Show where drift reports, cloud change history and reviewed definitions are stored. Explain how an operator compares them without exposing sensitive state.
Include how to identify the service owner for a resource. A technically detailed report is difficult to act on if nobody can decide which state is intended.
Give the receiving team a supported read-only diagnostic route and a separate controlled mutation path.
Hand over emergency reconciliation
Describe how urgent console changes are recorded and brought back into the maintained definition. Include who decides whether to retain or reverse them.
Link an example that shows the complete sequence from incident change to reviewed plan and verification. This makes the procedure more usable than a general rule against manual edits.
Keep temporary changes and unresolved drift visible across shifts, with a completion condition and owner.
Test independent operation
Ask the receiving team to investigate a safe synthetic difference and explain which controller should own the final value. Then have them prepare the intended reconciliation through the normal process.
Observe gaps in access, documentation and plan interpretation. Resolve those gaps before relying on the team during a production incident.
Assign responsibility for reviewing ownership when new automation is introduced. Infrastructure remains reproducible only when every controller participates in a coherent model, and that model needs maintenance as the platform evolves.
Primary sources
HashiCorp: managing resource driftReferences checked 11 September 2026.