How to initialize terraform variables in Azure DevOps pipeline

I have the following variables defined in my terraform variables file:

variable "environment" {
  type = string
}
variable "resource_group" {
  type = object({
    name        = string
    location    = string
  })
}

I initialize them in terraform.tfvars files per the following code snippet:

environment = "Dev"
resource_group = {
  name = "mytestresourcegroup"
  location = "Canada Central"
}

I have a build pipeline in Azure DevOps that runs my terraform implementation and creates the azure resources successfully. This is great, but it's not a scalable solution to create the resources in another environment e.g. production.

How can I initialize environment and resource_group variables' values int he Azure DevOps pipeline to something else so that for e.g. environment becomes "prod" or "uat", etc.? More importantly the resource_group is a more complex object and the question is how to initialize the name and the location.

One way to achieve this is to have a separate tf.vars file for each environment whose name will match you ADO pipeline environments. Once this is configured the ADO environment dev, uat, prd will be passed through your ADO pipeline to your terraform step via parameter substitution and map to the correct environment variable file.

I'd recommend a separate tf.vars file per environment as opposed to one that is overwritten as this will create a terraform dependency on your ADO pipelines. By creating a separate .tfvars for each environment then your terraform is more agnostic from your build/deployment process.

Here is a blog post detailing in depth how to achieve this and provides sample code.