Hashicorp Configuration Language
External
Internal
Overview
HCL (HashiCorp Configuration Language) is a specialized language for building infrastructure-as-code. It is not a fully featured programming language, so it requires understanding specific patterns to perform certain tasks, like conditional resources. The language supports re-usable modules.
Comments
# This is a comment
Block
Terraform Block
terraform {
required_version = ">= 0.12"
}
Backend Configuration
Backends are configured in Terraform files in the terraform block. Only one backend may be specified and the configuration may not contain interpolations.
terraform {
backend "local" {
path = ".../my-state.tfstate"
}
}
Resource Block
Declares a resource.
Data Block
Declares a data source.
data "data-source-name" "local-name" { ... }
data "aws_ami" "example" {
most_recent = true
owners = ["self", "0000000"]
filter {
name = "name"
values = ["consul-ubuntu-*"]
}
tags = {
Name = "app-server"
Tested = "true"
}
}
data "aws_eks_cluster" "cluster" {
name = module.eks.cluster_id
}
Data Sources
terraform_remote_state
This data source retrieves remote state from a Terraform backend.
Module Block
Represents a module call, referring to a dependency module that is going to be provisioned within the context of the current module.
module "some-module" {
source = "terraform-aws-modules/eks/aws"
version = "~> 9.0"
cluster_name = local.cluster_name
vpc_id = var.aws_vpc_id
...
}
More more details:
References to Named Values or Interpolation Expression
Resource elements can be used by other resources via interpolation expressions:
resource "something" "something_else" {
some_id = "something-${aws_vpc.main.id}"
...
}
Terraform 0.11 and earlier required all non-constant expressions to be provided via interpolation syntax ${...}, but this pattern is now deprecated in templates that consist entirely of a single interpolation sequence:
resource "aws_subnet" "something" {
ami = var.k8s_node_ami_id
...
}
Template interpolation syntax is still used to construct strings from expressions when the template includes multiple interpolation sequences or a mixture of literal strings and interpolations
Input Variable
Input variables are conventionally declared in the variables.tf file of the module:
variable "region" {
description = "This is the region"
type = string
default = "us-east-1"
}
Each variable can have a default value specified with "default". "default" may be "null". If there is no "default" declaration, Terraform will signal an error when terraform apply is executed, and it will initiate an interactive CLI that will inquire for the value of the missing variable:
var.environment_name The name of the environment Enter a value:
variable "masters" {
description = "The number of master nodes"
type = number
default = 1
}
A default value of 0 for the case where that input variable specifies a count of some sort, is valid. For a module that creates EC2 instances, specifying 0 for the default count value simply does not create any instance of that kind.
Variables can be used via interpolation expression, prefixed with “var.”
To use the variable:
...
Name = "${var.playground_name}-playground-vpc"
...
Variables can be assigned in multiples way, and this is the descending order of precedence:
- command-line flags:
-var = region=something)
It will complain if single quotes are used, they will be interpreted as part of the variable name. - from a file (.tfvars, -var-file=…) terraform.tfvars or *.auto.tfvars in the current directory are automatically loaded. Multiple -var-file can be used.
- From environment variables: TF_VAR_name. This can only be used for string variables.
- UI Input
- Variable defaults (“default” keyword).
Input Lists
variable "security_group_ids" {
description = "The list of security groups"
type = list
default = ["sg-a41f9d51704199e97"]
}
Input Maps
Output Variable
Output variables are a way to organize data to be easily queried and shown back to the Terraform user. Terraform usually stores hundred or thousands of attribute values, but only a few of those are important. Output variables have several uses: a child module can use outputs to expose a subset of its resource attributes to a parent module, a root module ca use outputs to print certain values in the CLI output after running terraform apply
and when using remote state, module outputs can be accessed by other configurations via a terraform_remote_state
data source.
"Output value" and "output" are semantically equivalent.
An output is declared as:
output "bastion_public_ip" {
value = aws_instance.bastion.public_ip
description = "The public IP address of the bastion"
}
Outputs can be delegated to other teams via remote state.
Also see:
Output Lists
output "master_node_private_ips" {
value = aws_instance.master-node.*.private_ip
description = "The private IP of the master nodes"
}
Result:
Outputs:
master_node_private_ips = [
"10.1.13.51",
]
worker_node_private_ips = [
"10.1.13.153",
"10.1.13.61",
]
Output Maps
Local Values
A local value assigns a name to an expression, allowing it to be used multiple times within a module without repeating it.
TODO: https://www.terraform.io/docs/configuration/locals.html
locals {
cluster_name = var.cluster_name != null ? var.cluster_name : var.environment_name
elb_tag = "Key=kubernetes.io/cluster/${local.cluster_name},Value=shared"
asg_enabled = length([for wg in var.worker_groups: wg.name if wg.autoscaling_enabled]) != 0
}
Functions
Function Categories
String
substr
...
cidr_block = "${substr("${var.my_cidr_block}", 0, 5)}.1.0/24"
...
IP Network
cidrsubnet
The following turns "10.10.0.0/16" into its first /24 subnet "10.10.1.0/24"
...
cidr_block = "${cidrsubnet("${var.vpc_cidr_block}", 8, 1)}"
...