Hello,

Context

We created our first project on DO called “A”. Now, we want to create another project called “Dev”, where we wish to deploy our development environment with a few resources (droplets, load balancer, k8s, etc.) using Terraform.

I already read all the articles on https://www.digitalocean.com/community/tutorial_series/how-to-manage-infrastructure-with-terraform and none specify how to resolve my problem.

Problem

The problem is that every time I terraform apply the resource ends up in project “A” (it is not only with the DO Spaces).

Currently

I want to use TF to provision a droplet test and publish our terraform.tfstate using a DO Spaces as the backend.

Firstly, I created (using the GUI) the DO Spaces called tf-state under “Dev” then executed TF commands. I discovered that is necessary to create the DO Spaces beforehand terraform init, otherwise it won’t work.

Code

So, I proceed to execute my TF folder, like so:

# Provider
terraform {
  required_providers {
    digitalocean = {
      source  = "digitalocean/digitalocean"
      version = "2.7.0"
    }
  }
  required_version = ">=0.14.9"
}

provider "digitalocean" {
  token = var.do_token

  spaces_access_id  = var.access_id
  spaces_secret_key = var.secret_key
}

# Project
resource "digitalocean_project" "dev" {
  name        = "dev"
  description = "Development details..."
  purpose     = "Service or API"
  environment = "Development"
  resources = [
    digitalocean_droplet.test.urn,
    digitalocean_spaces_bucket.tf-state.urn
  ]
}

# Backend
terraform {
  backend "s3" {
    endpoint                    = "fra1.digitaloceanspaces.com/"
    region                      = "us-west-1"
    key                         = "terraform.tfstate"
    bucket                      = "tf-state"
    skip_credentials_validation = true
    skip_metadata_api_check     = true
  }
}

# Bucket
resource "digitalocean_spaces_bucket" "tf-state" {
  name   = "tf-state"
  region = var.region
}

# Droplet
resource "digitalocean_droplet" "test" {
  name   = "test"
  size   = "s-1vcpu-1gb"
  image  = var.image
  region = var.region
}

Everything goes well when running:

terraform init \
-backend-config="access_key=$SPACES_ACCESS_KEY_ID" \
-backend-config="secret_key=$SPACES_SECRET_ACCESS_KEY"

But, when running terraform apply I get a message:

Error: Error creating Spaces bucket: BucketAlreadyExists: Bucket already exists
        status code: 409, request id: , host id: 

A solution is to comment # Bucket and the called line in the resource:

# Project
resource "digitalocean_project" "dev" {
  name        = "dev"
  description = "Development details..."
  purpose     = "Service or API"
  environment = "Development"
  resources = [
    digitalocean_droplet.test.urn,
    # digitalocean_spaces_bucket.tf-state.urn <= commented line
  ]
}

# Commented block
# # Bucket
# resource "digitalocean_spaces_bucket" "tf-state" {
#   name   = "tf-state"
#   region = var.region
# }

By doing this, all works, but the DO Spaces tf-state is transfered to the project “A”.

Question 1

It is quite cumbersome to track all the resources in "digitalocean_project" "dev" that are related to “Dev”. Is it possible to instruct TF to apply by default the provisioning to a specific project, like “Dev” (i.e. create another token for the new project)?

Question 2

How to create a DO Spaces bound to a specific project?

Any help is greatly appreciated.

Thanks,