# Terraform effective fields and auto-scaling drift > Source-anchored research for Terraform effective fields and auto-scaling drift. Run status: COMPLETED. Parent: [Atlas storage auto-scaling and disk reclaim via compact](https://llms-explorer.com/tree/atlas-storage-auto-scaling-and-disk-reclaim-via-compact/) · 3 facets · 12 facts · page: https://llms-explorer.com/tree/terraform-effective-fields-and-auto-scaling-drift/ ## How-to and procedures - Enable use_effective_fields and remove spec ignore_changes together, without unrelated edits. Restore previously removed analytics/read-only blocks in a prior apply if validation requires it. (low confidence) (verified as of 2026-10-30) — [source](https://raw.githubusercontent.com/mongodb/terraform-provider-mongodbatlas/master/docs/resources/advanced_cluster.md) ## Problems, failure modes and limitations - Historical first-person reports document the failure mode: a 2022 Terraform plan proposed disk shrinkage from 21 GB to configured 15 GB, and a 2023 Pulumi report describes reverting an auto-scaled M30 to configured M10. These reports establish historical reconciliation risk, not a current effective-fields defect. (medium confidence) — [source](https://stackoverflow.com/questions/71815826/terraform-mongodbatlas-disk-size-scale-down-unexpected) ## Facts and statements - For supported dedicated clusters, use_effective_fields separates configured sizing from effective provisioned sizing, preventing auto-scaling plan drift without spec ignore_changes blocks. (medium confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/cluster-autoscaling-operations.md) - Effective fields support dedicated M10+ clusters, exclude Flex clusters, and do not apply to search nodes. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/cluster-autoscaling-operations.md) - For electable and read-only specs, either compute or storage auto-scaling preserves instance size, disk size and disk IOPS independently of configured values; the Atlas table lists only instance size for analytics specs. (low confidence) (verified as of 2026-10-14) — [source](https://www.mongodb.com/docs/atlas/cluster-autoscaling-operations.md) - Read actual hardware through data-source effective_electable_specs, effective_read_only_specs and effective_analytics_specs; ordinary spec attributes describe configuration intent. (medium confidence) (verified as of 2026-10-30) — [source](https://registry.terraform.io/providers/mongodb/mongodbatlas/latest/docs/resources/advanced_cluster) - With effective fields and auto-scaling enabled, configured sizing changes do not affect live hardware. Disable compute and disk auto-scaling, apply desired specs, then re-enable scaling. (low confidence) (verified as of 2026-10-30) — [source](https://raw.githubusercontent.com/mongodb/terraform-provider-mongodbatlas/master/docs/resources/advanced_cluster.md) - Before disabling disk auto-scaling, nonzero advanced_configuration.oplog_min_retention_hours must be set to zero in a prior apply; otherwise HTTP 400 OPLOG_MIN_RETENTION_HOURS_NO_DISK_AUTO_SCALING can occur. (low confidence) (verified as of 2026-10-30) — [source](https://raw.githubusercontent.com/mongodb/terraform-provider-mongodbatlas/master/docs/resources/advanced_cluster.md) - For multiple replication_specs entries, the provider currently documents use_effective_fields as ineffective and advises against enabling it. (low confidence) (verified as of 2026-10-30) — [source](https://raw.githubusercontent.com/mongodb/terraform-provider-mongodbatlas/master/docs/resources/advanced_cluster.md) - Without effective fields, explicit ignore_changes protects auto-scaled sizing from reconciliation. Terraform uses ignored attributes at creation but ignores them for update planning. (medium confidence) — [source](https://registry.terraform.io/providers/mongodb/mongodbatlas/latest/docs/resources/advanced_cluster) - The changelog dates effective-fields resource support and effective-spec data-source outputs to provider 2.3.0, released December 9, 2025. (low confidence) — [source](https://raw.githubusercontent.com/mongodb/terraform-provider-mongodbatlas/master/CHANGELOG.md) - A 2023 provider issue records that static ignore_changes inside reusable modules blocks intentional instance-size updates in environments without auto-scaling. HashiCorp confirms ignored attributes are excluded from update planning. (medium confidence) — [source](https://github.com/mongodb/terraform-provider-mongodbatlas/issues/1299)