I have created s3 bucket using terraform. I am using random id for uniqueness.

resource "random_id" "s3_suffix-example" {
  byte_length = 4
}

resource "aws_s3_bucket" "s3_suffix-example-bucket" {
  bucket = "s3_suffix-example-bucket-${lower(random_id.s3_suffix-dev.id)}"

  tags = {
    Name  = "s3_suffix-example-bucket-${lower(random_id.s3_suffix-dev.id)}"
    Owner = "[email protected]"
  }
}

I found two problems with this approach though-

1- How do i modifying previously creating S3 bucket? e.g. I would like to add a policy to the s3 bucket created in first run. The first time tf apply ran it created bucket- s3_suffix-example-bucket-8hvg1g. When I run tf apply second time Terraform creates another bucket s3_suffix-example-bucket-rhwert. How do modify the bucket that was created in first run i.e.s3_suffix-example-bucket-8hvg1g?

2- I saw this is leaving behind dangling S3 buckets, if tf apply is run multiple times.

How can I solve these two problems?

CodePudding user response：

I don't know what you are doing to cause this behavior, but the behavior you're describing is not how this tool works. Consider the following:

resource "random_id" "this" {
  byte_length = 1
}

output "id" {
  value = random_id.this.id
}

When one applies this, you get an output. For me id = "sQ". Then, if I apply again, id = "sQ". And again ad infinitum.

So if defining a bucket in the same state:

resource "aws_s3_bucket" "this" {
  bucket = "my-bucket-${random_id.this.id}"
}

That bucket now has the name my-bucket-sQ. And when applied again, No changes. Your infrastructure matches the configuration.

And you can still modify as you see fit, or attach resources to it.

resource "aws_s3_bucket_acl" "this" {
  bucket = aws_s3_bucket.this.id
  acl    = "private"
}

As an aside, you should consider using the bucket_prefix argument if your goal is bucket name uniqueness. That is exactly for what it is designed, and doesn't require any other consideration.

resource "aws_s3_bucket" "this" {
  bucket_prefix = "my-bucket"
}