Terraform inconsistent behavior for aws_rds_cluster changes

[hashibot]

This issue was originally opened by @anosulchik as hashicorp/terraform#15715. It was migrated here as a result of the provider split. The original body of the issue is below.

---

We noticed that in case both availability_zones and db_subnet_group_name are specified for aws_rds_cluster then terraform tries to re-create cluster every time when we run terraform apply, even if no changes into resource config were made. Looks like we should allow only one parameter at a time.

    apply_immediately:                "" => "<computed>"
    availability_zones.#:             "3" => "2" (forces new resource)
    availability_zones.1305112097:    "us-east-1b" => "us-east-1b"
    availability_zones.3569565595:    "us-east-1a" => "us-east-1a"
    availability_zones.986537655:     "us-east-1c" => "" (forces new resource)
    backup_retention_period:          "35" => "35"
    cluster_identifier:               "tf-aurora-ecs-cluster-prod-cluster" => "tf-aurora-ecs-cluster-prod-cluster"
    cluster_identifier_prefix:        "" => "<computed>"
    cluster_members.#:                "1" => "<computed>"
    cluster_resource_id:              "cluster-xxx" => "<computed>"
    database_name:                    "" => "<computed>"
    db_cluster_parameter_group_name:  "tf-aurora-ecs-cluster-prod-cluster-params" => "tf-aurora-ecs-cluster-prod-cluster-params"
    db_subnet_group_name:             "shared-cluster-rds-prod" => "shared-cluster-rds-prod"
    endpoint:                         "tf-aurora-ecs-cluster-prod-cluster.cluster-cexkw116s3ng.us-east-1.rds.amazonaws.com" => "<computed>"
    engine:                           "aurora" => "<computed>"
    kms_key_id:                       "arn:aws:kms:us-east-1:xxx:key/xxx" => "<computed>"
    master_password:                  "<sensitive>" => "<sensitive>" (attribute changed)
    master_username:                  "root" => "root"
    port:                             "3306" => "<computed>"
    preferred_backup_window:          "01:00-03:30" => "01:00-03:30"
    preferred_maintenance_window:     "sat:04:00-sat:04:30" => "sat:04:00-sat:04:30"
    reader_endpoint:                  "tf-aurora-ecs-cluster-prod-cluster.cluster-ro-xxx.us-east-1.rds.amazonaws.com" => "<computed>"
    skip_final_snapshot:              "true" => "true"
    storage_encrypted:                "true" => "true"
    vpc_security_group_ids.#:         "1" => "1"
    vpc_security_group_ids.859373895: "sg-xxx" => "sg-xxx"```

Thank you!

[eschwartz]

I know this is an old issue, but does anyone have any information about what's going on here? Does RDS require that you always provide 3 availability zones?

[rottenbytes]

@eschwartz : You can circumvent this by not specifying the AZs but only the subnet group : since you will create that subnet group with subnet only in the AZs you selected it will use only those (I do it for 2 AZs im my staging environment). Small precision if relevant : I use aurora, but it should impact.

[bflad]

Hi everyone 👋 this has been answered in #1111 - please see that issue for more information.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!