Terraform tries to re-create RDS instance when nothing has changed-CodePudding

I have the following Terraform configuration that is used to create a cluster with two instances:

resource "aws_rds_cluster" "aurora" {
  storage_encrypted  = true
  cluster_identifier = var.cluster_identifier
  engine             = "aurora-postgresql"
  engine_mode        = "provisioned"
  engine_version     = "13.6"
  database_name      = var.database_name
  master_username    = "test"
  master_password    = var.database_password
  availability_zones = ["ap-southeast-2a", "ap-southeast-2b"]

  db_subnet_group_name = var.db_subnet_group_name
  serverlessv2_scaling_configuration {
    max_capacity = 1.0
    min_capacity = 0.5
  }

  tags = {
    Name        = "${var.prefix}-${var.environment}-rds-cluster"
    Environment = "${var.prefix}-${var.environment}"
  }
  vpc_security_group_ids = var.aurora_security_group_id
  skip_final_snapshot    = true
}

resource "aws_rds_cluster_instance" "aurora" {
  count                = 2
  availability_zone    = var.availability_zones[count.index]
  cluster_identifier   = aws_rds_cluster.aurora.id
  instance_class       = "db.serverless"
  engine               = aws_rds_cluster.aurora.engine
  engine_version       = aws_rds_cluster.aurora.engine_version
  db_subnet_group_name = var.db_subnet_group_name
  publicly_accessible  = true
}

If I make literally no changes to my configuration, and run terraform plan, I receive the output below:

module.potentium_databases.module.potentium_rds_cluster.aws_rds_cluster_instance.aurora[1] must be replaced
-/  resource "aws_rds_cluster_instance" "aurora" {
        apply_immediately                     = (known after apply)
      ~ arn                                   = "arn:aws:rds:ap-southeast-2:749732509682:db:tf-20220706042316120800000001" -> (known after apply)
      ~ ca_cert_identifier                    = "rds-ca-2019" -> (known after apply)
      ~ cluster_identifier                    = "potentium-cluster" -> (known after apply) # forces replacement
      ~ db_parameter_group_name               = "default.aurora-postgresql13" -> (known after apply)
      ~ dbi_resource_id                       = "db-5AH6GR5KJNW4IXQ2BSGNPLL4FM" -> (known after apply)
      ~ endpoint                              = "tf-20220706042316120800000001.cv6x1exxvfdc.ap-southeast-2.rds.amazonaws.com" -> (known after apply)
      ~ engine_version_actual                 = "13.6" -> (known after apply)
      ~ id                                    = "tf-20220706042316120800000001" -> (known after apply)
      ~ identifier                            = "tf-20220706042316120800000001" -> (known after apply)
        identifier_prefix                     = (known after apply)
      ~ kms_key_id                            = "arn:aws:kms:ap-southeast-2:749732509682:key/a3f87bb9-f0b4-44a4-8677-bac5f0bb1546" -> (known after apply)
        monitoring_role_arn                   = (known after apply)
      ~ performance_insights_enabled          = false -> (known after apply)
        performance_insights_kms_key_id       = (known after apply)
      ~ performance_insights_retention_period = 0 -> (known after apply)
      ~ port                                  = 5432 -> (known after apply)
      ~ preferred_backup_window               = "13:51-14:21" -> (known after apply)
      ~ preferred_maintenance_window          = "thu:15:39-thu:16:09" -> (known after apply)
      ~ storage_encrypted                     = true -> (known after apply)
      - tags                                  = {} -> null
      ~ tags_all                              = {} -> (known after apply)
      ~ writer                                = false -> (known after apply)
        # (10 unchanged attributes hidden)
    }

Can anyone explain why Terraform thinks this resource needs to be recreated even if nothing has changed? It is causing me grief due to how long it takes to actually re-create the instances.

CodePudding user response：

It appears my issue is that I was only specifying 2 Availability Zones instead of 3. Im assuming because Terraform/AWS is left to decide the third AZ, it must perform a re-create as it does not know what to use.

CodePudding user response：

As I know, the re-creating issue is usually caused by the changed attribute, which is shown with #forces replacement. In your case:

~ cluster_identifier = "potentium-cluster" -> (known after apply) # forces replacement

Please double-check it.