Lesson 3 of 7 in Infrastructure as Code with Terraform and OpenTofu

Resources, variables and outputs

Core10 min

A resource is one thing the tool manages. A variable is an input you can change without editing the resource. An output is a value the configuration reports after apply. References between them build the dependency graph that decides the order of work.

Draft. This lesson is written and waiting for a person to run every command and review it. It is safe to learn from, but check the official documentation for anything you depend on.

Read these first: The workflow: init, plan, apply, destroy

The problem this solves

A configuration with every value written in place works once. As soon as you need a second copy with different names or sizes, you are copy-pasting files and the copies drift apart.

Variables let one configuration produce different results from different inputs. Outputs hand useful results, such as an address or a file path, back to you or to another configuration.

How it works

Values flow in through variables and out through outputs
Values flow in through variables and out through outputsYou supply a value, it becomes a variable, a resource uses it, another resource references the first, and an output reports the result.1 you set it2 used as input3 referencedby4 reportedYour value-var or tfvars fileVariablevar.greetingrandom_petresourcelocal_fileresourceOutputshown after apply
  • People and clients
  • Control
  • Runs code
  • Other

You supply a value, it becomes a variable, a resource uses it, another resource references the first, and an output reports the result.

  1. Your value to Variable: 1 you set it
  2. Variable to local_file: 2 used as input
  3. random_pet to local_file: 3 referenced by
  4. local_file to Output: 4 reported

Reading the diagram from the right: you pass a value on the command line or in a terraform.tfvars file, and it lands in a variable declared with a variable block. The local_file uses var.greeting and also refers to random_pet.name.id.

Every reference like random_pet.name.id is a dependency. The tool builds a graph from them, creates the pet first, then the file, and can create unrelated resources in parallel. You rarely order things by hand.

Variables can have a type, a default and a description. A variable marked sensitive = true is hidden from plan and output text. It is still stored in state, which the next concept covers.

Try it yourself

Same free providers as before, in a fresh folder.

  1. 1
    Write config with a variable and an output
    mkdir -p ~/iac-lab/vars && cd ~/iac-lab/vars
    cat > main.tf <<"EOF"
    terraform {
      required_providers {
        local  = { source = "hashicorp/local" }
        random = { source = "hashicorp/random" }
      }
    }
    
    variable "greeting" {
      type        = string
      description = "Word at the start of the file"
      default     = "Hello"
    }
    
    resource "random_pet" "name" {
      length = 2
    }
    
    resource "local_file" "note" {
      filename = "note.txt"
      content  = format("%s from %s", var.greeting, random_pet.name.id)
    }
    
    output "who" {
      value = random_pet.name.id
    }
    EOF
    terraform init
  2. 2
    Apply with the default
    terraform apply -auto-approve
    cat note.txt; echo

    -auto-approve skips the yes prompt. Use it only in a sandbox like this one. The output block prints who = followed by the pet name.

  3. 3
    Override the variable
    terraform apply -auto-approve -var 'greeting=Howdy'
    cat note.txt; echo

    Only the file changes. The pet name is kept because nothing about it changed, which you can confirm in the plan.

  4. 4
    Put inputs in a file
    echo 'greeting = "Namaste"' > terraform.tfvars
    terraform plan

    terraform.tfvars is read automatically. The plan shows the content changing from Howdy to Namaste. Command-line -var beats the tfvars file when both are set.

  5. 5
    Read outputs later, then clean up
    terraform output who
    terraform destroy -auto-approve

    terraform output reads from state, so it works any time after apply.

What breaks, and where to look

  • Error: No value for required variable (or an interactive prompt asking for var.something).

    WHY IT HAPPENS
    The variable has no default and no value was supplied through -var, a tfvars file or a TF_VAR_ environment variable. In automation there is no one to answer the prompt.
    CHECK FIRST
    Run terraform plan -var 'name=value' or add the value to terraform.tfvars. Check the variable name spelling in the error.
  • Error: Unsupported argument. An argument named "contents" is not expected here.

    WHY IT HAPPENS
    A misspelled argument name. Each resource type accepts a fixed set of arguments, and the wrong name is rejected before anything runs.
    CHECK FIRST
    Read the resource's page in the provider documentation, for example the local_file page, and compare argument names.
  • A secret you marked sensitive is still readable somewhere.

    WHY IT HAPPENS
    sensitive = true only hides the value from console output. The value is still written to the state file in plain text.
    CHECK FIRST
    Open terraform.tfstate and search for the value. Keep state in a protected location and prefer reading secrets at runtime from a secrets service.

When not to use this

  • Do not turn every number into a variable. Variables you never change make a configuration harder to read. Add one when a value actually differs between uses.
  • Putting secrets in terraform.tfvars and committing it exposes them. Keep such files out of Git or supply values through environment variables.

Check yourself

How does the tool decide which resource to create first?

From references between resources. If one uses another's attribute, the referenced one is created first.

Which wins when a value is given both with -var and in terraform.tfvars?

The -var command-line value wins.

Does sensitive = true keep a value out of the state file?

No. It only hides the value from displayed output. The state file still holds it in plain text.

Lesson 65 of 92 in the DevOps path

Search

Search cheat sheets and projects