TFC CLI-driven Run

Expected Outcome

In this lab, we will invoke a Terraform run on TFC via your local Terraform client (Terraform OSS binary). This is also referred to as the CLI-driven run.

How To

1. In the TFC UI, create a new Workspace. Click CLI-Driven Workflow . Name the Workspace something unique, such as ###-tfcli-test-run.

2. Still in the TFC UI, within the Variables section of your newly created TFC Workspace, create the four necessary environment variables for TFC to authenticate to Azure (and mark them as sensitive):

ARM_SUBSCRIPTION_ID
ARM_CLIENT_ID
ARM_CLIENT_SECRET
ARM_TENANT_ID

3. Create a new directory locally on your workstation and create a main.tf within it that will provision an Azure Resource Group (or any other basic Azure resource you feel like provisioning - it doesn't really matter what is provisioned for this example):

main.tf

provider "azurerm" {
  features {}
}

resource "azurerm_resource_group" "rg" {
  name     = "###-temp-rg"
  location = "eastus"

  tags = {
    terraform = "true"
  }
}

Note: fill in your initals for ### so you don't overlap / cause duplicates with your classmates.

1. Within the same local working directory, create a backend.tf file to tell your local Terraform client how to reach your TFE instance:

backend.tf

terraform {
  backend "remote" {
    hostname     = "app.terraform.io"
    organization = "<my-tfe-org-name>"

    workspaces {
      name = "###-tfecli-test-run"
    }
  }
}

where organization is your specific TFC Organization name, and name within the workspaces block is your newly created Workspace name.

1. terraform init
2. terraform plan
3. terraform apply - refresh the TFC UI and look for the running apply within your TFC Workspace

Summary

The concepts with the CLI-driven run method are that we need a backend.tf file so the local Terraform client knows where to make its API calls against TFC, and we also need a TFC API token so we can properly authenticate to our TFC instance. It is also important to note that we did not connect our Workspace to a VCS repo. This is because the Terraform client takes care of compressing and sending the code to the TFC workspace via the TFC API.

This method can be run locally as we just demonstrated, or it can be executed from a build script/ CI pipeline.