Amazon Aurora clusters should have backtracking enabled

How to fix it

• Replaces the resource: Backtrack is set only at creation, and in Terraform backtrack_window plans an in-place update that cannot enable it: clone or restore with Backtrack on, move traffic, then delete this cluster.
• Can break clients: Cross-Region read replicas cannot be created from a backtrack-enabled cluster, though binary log replication still works.

In the console

RDS → Databases → Create clone or Create database → Backtrack

Enable Backtrack

Selected

Target Backtrack window

How far back you may need to rewind, up to 72 hours

With the AWS CLI

aws rds restore-db-cluster-to-point-in-time --db-cluster-identifier <new-db-cluster-identifier> --source-db-cluster-identifier <source-db-cluster-identifier> --restore-type copy-on-write --use-latest-restorable-time --backtrack-window <seconds> --region <region>

With Terraform

resource "aws_rds_cluster" "this" {
  engine           = "aurora-mysql"
  backtrack_window = var.backtrack_window_seconds
}