Terraform scripts can become quite long and involved, especially when you’re builing out larger projects. I recently worked on a project which needed multiple subnets. Each subnet needed a different cidr_block, but otherwise they were pretty much the same.
Rather than duplicating the same code block 6 times to set up each of the subnets I opted for keeping the code DRY and looping over a list of objects to get each unique cidr_block. This keeps the code clean and uncluttered, and the subnets are all configured in once place, in the variables file (terraform.tfvars).
We’ll need three pieces of code for this method to work:
variable to state which unique values can be defined for each item we createterraform.tfvars fileterraform.tfvars file to create the multiple resourcesThe first thing to do is create a variable which can be used to set up the unique values for each resource. In the example of subnets in a VPC, the main unique value to each subnet is the cidr_block. This is simply a string value representing an IP address or IP range.
In your variables.tf file, create a variable with the type of list and inside that list, define an object. Within this object is where we will specify the unique elements:
variable "subnets" {
type = list(object({
cidr_block = string
name_tag = string
}))
}variables.tf
Once this is done, we can input this data via the terraform.tfvars file. Here we’re adding the data for 4 subnets with different cidr_blocks and a tag to easily distinguish between each subnet once the resource is provisioned.
subnets = [
{
cidr_block = "10.0.2.0/24"
name_tag = "Private Subnet 1 - App Tier"
},
{
cidr_block = "10.0.3.0/24"
name_tag = "Private Subnet 2 - App Tier"
},
{
cidr_block = "10.0.4.0/24"
name_tag = "Private Subnet 3 - DB Tier"
},
{
cidr_block = "10.0.5.0/24"
name_tag = "Private Subnet 4 - DB Tier"
}
]terraform.tfvars
The last step is to write the actual resource block for the subnets. To loop over the list and create multiple resources, we can use the count meta-argument. count requires that we input an integer to represent the number of objects we want to create, so you would typically write something like count = 2 to create 2 of a particular resource.
We can provide this integer by finding out the length of our array, and this is done using the length() function, which returns the length of a list, map or string.
resource "aws_subnet" "subnets" {
count = length(var.subnets)
vpc_id = aws_vpc.main.id
cidr_block = var.subnets[count.index].cidr_block
tags = {
Name = var.subnets[count.index].name_tag
}
}In the above example:
count meta-argument by finding the length of the subnets list. In this case it is 4, because we have 4 unique objects.count.index increments by one, and the cidr_block is determined by getting the cidr_block from each object.Name tag for each resource, extracting the value from the name_tag of each object.This example will provision 4 subnets, but the same technique can be applied to creating multiple copies of any resource, and you can further expand upon this by adding more data into the objects in your variable. The resulting Terraform code is a lot smaller and easier to read - rather than 30+ lines of code to provision these resources, we’ve done it in 9 lines. The benefits of this increase the more “duplicates” you need to create.
It is worth keeping in mind though, that this might not be an appropriate method for every situation. If you were only creating 2 copies of a resource, this might be overkill rather than just writing the resource block for the 2 items. Equally, if you have lots of options for each object, it may become difficult to see what exactly will be created, and it is sometimes better to be verbose and clearly state your intentions for each unique resource block.