Terraform For each iteration

This is to explain Terraform for each looping technique. In this example, 3 buckets are created to demonstrate the looping idea.

In the first step, we will create the above 3 buckets starting from 0.

---

• Simple Example
• Count
• For each

---

Simple Example

This is a simple example¹ where you have only string interpolations in the local variables. In the main.tf file:

terraform {
  required_version = ">=0.12.0"
}

provider "aws" {
  version = "~> 2.0"
  region  = "ap-southeast-2"
  profile = "dev"
}

variable "bucket_owner" {
  type    = string
  default = "oj"

}


locals {
  bucket_prefix= "${data.aws_caller_identity.current.account_id}-${var.bucket_owner}-${local.bucket_resource}"
  bucket_resource = "bucket"    
}

resource "aws_s3_bucket" "bucket0" {
  bucket =  "${local.bucket_prefix}0"
}

resource "aws_s3_bucket" "bucket1" {
  bucket = "${local.bucket_prefix}1"
  tags = {
    "dependency" = aws_s3_bucket.bucket0.arn
  }
}

resource "aws_s3_bucket" "bucket2" {
  bucket = "${local.bucket_prefix}2"
  #Explicit
  depends_on = [aws_s3_bucket.bucket1]
}

data "aws_caller_identity" "current" {

}

data "aws_availability_zones" "available" {
  state = "available"
}

output "bucket1_info" {
  value = aws_s3_bucket.bucket1
}

output "bucket2_info" {
  value = aws_s3_bucket.bucket1
}

output "aws_caller_info" {
  value = data.aws_caller_identity.current
}

output "aws_availability_zones" {
  value = data.aws_availability_zones.available
}

output "aws_caller_identity_account_id" {
  value = data.aws_caller_identity.current
}

The bucket name is composed in line #19 under the local variables.

Count

You can use the count.index to start the base number. For example, if the count value is 3, the bucket is 1 to 2, not starting from zero.

terraform {
  required_version = ">=0.12.0"
}

provider "aws" {
  version = "~> 2.0"
  region  = "ap-southeast-2"
  profile = "dev"
}

variable "bucket_owner" {
  type    = string
  default = "oj"

}


locals {
  bucket_prefix= "${data.aws_caller_identity.current.account_id}-${var.bucket_owner}-${local.bucket_resource}"
  bucket_resource = "bucket"    
}

resource "aws_s3_bucket" "bucket" {
  count = 3
  bucket =  "${local.bucket_prefix}${count.index}"
}

# resource "aws_s3_bucket" "bucket1" {
#   bucket = "${local.bucket_prefix}1"
#   tags = {
#     "dependency" = aws_s3_bucket.bucket0.arn
#   }
# }

# resource "aws_s3_bucket" "bucket2" {
#   bucket = "${local.bucket_prefix}2"
#   #Explicit
#   depends_on = [aws_s3_bucket.bucket1]
# }

data "aws_caller_identity" "current" {

}

data "aws_availability_zones" "available" {
  state = "available"
}

output "bucket_info" {
  value = aws_s3_bucket.bucket
}

# output "bucket2_info" {
#   value = aws_s3_bucket.bucket2
# }

output "aws_caller_info" {
  value = data.aws_caller_identity.current
}

output "aws_availability_zones" {
  value = data.aws_availability_zones.available
}

output "aws_caller_identity_account_id" {
  value = data.aws_caller_identity.current
}

The above provision script is simplified as follows:

The magic count has been used in lines #24-25.

For each

Here,

terraform {
  required_version = ">=0.12.0"
}

provider "aws" {
  version = "~> 2.0"
  region  = "ap-southeast-2"
  profile = "dev"
}

variable "bucket_owner" {
  type    = string
  default = "oj"

}


locals {
  bucket_prefix= "${data.aws_caller_identity.current.account_id}-${var.bucket_owner}-${local.bucket_resource}"
  bucket_resource = "bucket"    
}

locals {
  buckets = {
    bucket0 = "0"
    bucket1 = "1"
    bucket2 = "2"
  }
}

resource "aws_s3_bucket" "bucket" {
  for_each = local.buckets
  bucket =  "${local.bucket_prefix}${each.value}"
}

# resource "aws_s3_bucket" "bucket1" {
#   bucket = "${local.bucket_prefix}1"
#   tags = {
#     "dependency" = aws_s3_bucket.bucket0.arn
#   }
# }

# resource "aws_s3_bucket" "bucket2" {
#   bucket = "${local.bucket_prefix}2"
#   #Explicit
#   depends_on = [aws_s3_bucket.bucket1]
# }

data "aws_caller_identity" "current" {

}

data "aws_availability_zones" "available" {
  state = "available"
}

output "bucket_info" {
  value = aws_s3_bucket.bucket
}

# output "bucket2_info" {
#   value = aws_s3_bucket.bucket2
# }

output "aws_caller_info" {
  value = data.aws_caller_identity.current
}

output "aws_availability_zones" {
  value = data.aws_availability_zones.available
}

output "aws_caller_identity_account_id" {
  value = data.aws_caller_identity.current
}

In the above code, the count has been replaced by the foreach lines 23-34 as follows:

You can change the above code’s map to list as follows:

locals {
  buckets = ["0","1","2"]
}

resource "aws_s3_bucket" "bucket" {
  for_each = toset(local.buckets)
  bucket =  "${local.bucket_prefix}${each.key}"
}

As shown in lines #1-3, only the list elements are created as buckets. Please note how line #6 has been changed to return set. In line #7 instead of each.value now we are using each.key.

1. Developing Infrastructure as Code with Terraform LiveLessons ↩