terraform import aws_rds_cluster_instance fails with unexpected EOF error

[kronos]

Terraform Version

Terraform v0.11.5
+ provider.aws v1.12.0

Affected Resource(s)

• aws_rds_cluster_instance

Debug Output

https://gist.github.com/kronos/228520eca6f64469753b41b14779e79c

Expected Behavior

Imported aws_rds_cluster_instance

Actual Behavior

module.rds.aws_rds_cluster_instance.cluster_instances: Importing from ID "postgres-development"...
module.rds.aws_rds_cluster_instance.cluster_instances: Import complete!
  Imported aws_rds_cluster_instance (ID: postgres-development)
module.rds.aws_rds_cluster_instance.cluster_instances: Refreshing state... (ID: postgres-development)

Error: module.rds.aws_rds_cluster_instance.cluster_instances (import id: postgres-development): 1 error(s) occurred:

* import module.rds.aws_rds_cluster_instance.cluster_instances result: postgres-development: aws_rds_cluster_instance.cluster_instances: unexpected EOF

Steps to Reproduce

1. Try to create a rds cluster with one instance with terraform (it will create aws_rds_cluster but fail to create instance with error below)

 InvalidParameterCombination: RDS does not support creating a DB instance with the following combination: DBInstanceClass=db.t2.micro, Engine=aurora-postgresql, EngineVersion=9.6.6, LicenseModel=postgresql-license. For supported combinations of instance class and database engine version, see the documentation.

2. Create it manually in amazon web interface

3. Try to import it with command

terraform import module.rds.aws_rds_cluster_instance.cluster_instances postgres-development

4. Just in case, a piece of my config:

resource "aws_rds_cluster" "postgres" {
  cluster_identifier     = "postgres-${var.env}"
  engine                 = "aurora-postgresql"
  database_name          = "equipage"
  port                   = "${var.port}"
  master_username        = "${var.username}"
  master_password        = "${var.password}"
  vpc_security_group_ids = ["${aws_security_group.postgres-sg.id}"]
  db_subnet_group_name   = "${aws_db_subnet_group.postgres-subnet.name}"
  skip_final_snapshot    = true

  lifecycle {
    ignore_changes = ["master_password"]
  }
}

resource "aws_rds_cluster_instance" "cluster_instances" {
  count                = "${var.instance_count}"
  identifier           = "postgres-instance-${var.env}"
  cluster_identifier   = "${aws_rds_cluster.postgres.id}"
  instance_class       = "${var.instance_class}"
  engine               = "aurora-postgresql"
  db_subnet_group_name = "${aws_db_subnet_group.postgres-subnet.name}"
}

[bflad]

Hi @kronos 👋 are you positive the new database instance you created was a member of the Aurora cluster? The debug output shows the cluster as having no members:

2018-03-28T21:54:57.069+0300 [DEBUG] plugin.terraform-provider-aws_v1.12.0_x4:         <DBClusterMembers/>

Non-Aurora database instances need to be managed by the aws_db_instance resource.

[kronos]

Hello, Brian. No, I'm negative, thank you, I understand my mistake.

[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!