Using terragrunt generate provider block causes conflicts with require providers block in module - terraform

I'm using Terragrunt with Terraform version 0.14.8.
My project uses mono repo structure as it is a project requirement to package Terragrunt files and Terraform modules together in a single package.
Folder structure:
project root:
├── environments
│   └── prd
│   ├── rds-cluster
│   │   └── terragrunt.hcl
│   └── terragrunt.hcl
└── modules
├── rds-cluster
│   ├── README.md
│   ├── main.tf
│   ├── output.tf
│   └── variables.tf
└── secretsmanager-secret
├── README.md
├── main.tf
├── output.tf
└── variables.tf
In prd/terragrunt.hcl I define the remote state block and the generate provider block.
remote_state {
backend = "s3"
...
}
generate "provider" {
path = "provider.tf"
if_exists = "overwrite_terragrunt"
contents = <<EOF
terraform {
required_providers {
aws = {
source = "hashicorp/aws"
version = "~> 3.0"
}
}
}
provider "aws" {
region = "ca-central-1"
}
EOF
}
In environments/prd/rds-cluster/terragrunt.hcl, I defined the following:
include {
path = find_in_parent_folders()
}
terraform {
source = "../../../modules//rds-cluster"
}
inputs = {
...
}
In modules/rds-cluster/main.tf, I defined the following:
terraform {
required_providers {
aws = {
source = "hashicorp/aws"
version = ">= 3.0"
}
}
}
// RDS related resources...
My problem is that when I try to run terragrunt plan under environments/prd/rds-cluster, I get the following error message:
Error: Duplicate required providers configuration
on provider.tf line 3, in terraform:
3: required_providers {
A module may have only one required providers configuration. The required
providers were previously configured at main.tf:2,3-21.
I can resolve this by declaring the version within the provider block as shown here. However, the version attribute in provider blocks has been deprecated in Terraform 0.13; Terraform recommends the use of the required_providers sub-block under terraform block instead.
Does anyone know what I need to do to use the new required_providers block for my aws provider?

As you've seen, Terraform expects each module to have only one definition of its required providers, which is intended to avoid a situation where it's unclear why Terraform is detecting a particular when the declarations are spread among multiple files.
However, to support this sort of piecemeal code generation use-case Terraform has an advanced feature called Override Files which allows you to explicitly mark certain files for a different mode of processing where they selectively override particular definitions from other files, rather than creating entirely new definitions.
The details of this mechanism depend on which block type you're overriding, but the section on Merging terraform blocks` discusses the behavior relevant to your particular situation:
If the required_providers argument is set, its value is merged on an element-by-element basis, which allows an override block to adjust the constraint for a single provider without affecting the constraints for other providers.
In both the required_version and required_providers settings, each override constraint entirely replaces the constraints for the same component in the original block. If both the base block and the override block both set required_version then the constraints in the base block are entirely ignored.
The practical implication of the above is that if you have an override file with a required_providers block that includes an entry for the AWS provider then Terraform will treat it as a full replacement for any similar entry already present in a non-override file, but it won't affect other provider requirements entries which do not appear in the override file at all.
Putting all of this together, you should be able to get the result you were looking for by asking Terragrunt to name this generated file provider_override.tf instead of just provider.tf, which will then activate the override file processing behavior and thus allow this generated file to override any existing definition of AWS provider requirements, while allowing the configurations to retain any other provider requirements they might also be defining.

Related

Terraform not declaring tfvars

I am new to Terraform and I am writing a script. Following is my directory structure
folder
---.terraform
---..terraform.lock.hcl
---main.tf
---terraform.tfvars
---variables.tf
Following is my content on terraform.tfvars.
environment = "development"
Following is my content on main.tf.
tags = {
environment = var.environment
}
But the values are not updating. Following is the error:
╷
│ Warning: Value for undeclared variable
│
│ The root module does not declare a variable named "environment" but a value was found in file "terraform.tfvars". If you meant to use this value, add a "variable" block to the configuration.
│
│ To silence these warnings, use TF_VAR_... environment variables to provide certain "global" settings to all configurations in your organization. To reduce the verbosity of these warnings, use the
│ -compact-warnings option.
╵
╷
│ Warning: Value for undeclared variable
│
│ The root module does not declare a variable named "admin_username" but a value was found in file "terraform.tfvars". If you meant to use this value, add a "variable" block to the configuration.
│
│ To silence these warnings, use TF_VAR_... environment variables to provide certain "global" settings to all configurations in your organization. To reduce the verbosity of these warnings, use the
│ -compact-warnings option.
╵
╷
│ Warning: Values for undeclared variables
│
│ In addition to the other similar warnings shown, 1 other variable(s) defined without being declared.
╵
╷
│ Error: Reference to undeclared input variable
│
│ on main.tf line 22, in resource "azurerm_resource_group" "tf_example_rg":
│ 22: environment = var.environment
│
│ An input variable with the name "environment" has not been declared. This variable can be declared with a variable "environment" {} block.
As I am using terraform.tfvars I don't need to give the filename on CLI. I think I am doing everything right but it's yet not working.
You have to actually declare your variable using variable block. For example:
variable "environment" {}
If you have such declarations, you have to double check the spelling and locations of them.
#AunZaidi , As stated in the error messages terraform can not find the defined variables.
The root module does not declare a variable named "environment" but a value was found in file "terraform.tfvars". If you meant to use this value, add a "variable" block to the configuration.
I would recommend you to take a look at the terraform-azure-tutorials to get acquainted with the basics.
you can solve your issue by just defining a new variable using syntax
variable "environment" {
type = string
description = "(optional) Environment for the deployment"
}
Refer to https://developer.hashicorp.com/terraform/language/values/variables#arguments for definitions of the arguments used in terraform variables.
Also one of the recommended practices is to use a dedicated file variables.tf for all the variables inputs required in your terraform code.

Terraform upgrade and multiple versions.tf, do child modules inherit the provider versions?

Not very experienced with Terraform. I upgraded my project from 12 to 13 and looking to upgrade it to 14 afterwards.
As the documentation specifies, I ran terraform 0.13upgrade and terraform 0.13upgrade module, my directory became like this:
terraform
├── module
│ ├── main.tf
│ └── versions.tf
├── main.tf
└── versions.tf
I moved my versions but the problem is that I specified them only in the root versions.tf:
terraform {
required_providers {
google = {
source = "hashicorp/google"
version = "~> 3.8.0"
}
}
in module/versions.tf I kept only:
terraform {
required_providers {
google = {
source = "hashicorp/google"
}
}
Do child modules inherit the version from the root (where they are imported) or does this mean my module will automatically run with a more recent provider version (3.64 I think)?
Should I simply remove module/versions.tf? (It's tiresome to have 2 versions to edit everytime).
Thanks!
For a given Terraform configuration (which includes both the root module and any other modules you might call), there can be only one version of each provider. Terraform recognizes that two providers are "the same" by them having the same source value after normalization, and in your examples here both are using hashicorp/google, which is short for registry.terraform.io/hashicorp/google, and so both of them need to be able to agree on a particular version of that provider to use.
Terraform handles version constraints from multiple providers by combining them together and trying to honor all of them together. In your examples here you've written no version argument in the child module, and this means "any version is allowed".
Terraform will therefore look for an available provider version that matches both the ~> 3.8.0 constraint and the implied "any version" constraint, and since the ~> 3.8.0 constraint is a proper subset of "any version" it effectively takes priority over the open constraint in the child module. This is not strictly "inheritance", but it happens to behave somewhat like it in this case because the child module is totally unconstrained.
A more interesting example would be if both of your modules specified different version constraints, which means we can see a more interesting effect of combining them. Let's pretend that your two modules had the following requirements instead:
terraform {
required_providers {
google = {
source = "hashicorp/google"
version = "~> 3.8.2"
}
}
}
terraform {
required_providers {
google = {
source = "hashicorp/google"
version = ">= 3.8.5"
}
}
}
In this situation neither of these constraints is a subset of the other, but they do have some overlap: all of the 3.8.x versions from 3.8.5 onwards are acceptable to both modules. Therefore Terraform will select the newest available version from that set.
If you write two modules that have conflicting version constraints then that would be an error:
terraform {
required_providers {
google = {
source = "hashicorp/google"
version = "~> 3.7.0"
}
}
}
terraform {
required_providers {
google = {
source = "hashicorp/google"
version = "~> 3.8.5"
}
}
}
There is no provider version that is both a 3.7.x release and a 3.8.x release at the same time, so no release can ever possibly match both of these constraints, and thus provider version selection will fail. It's for this reason that the Terraform documentation section Best Practices for Provider Versions advises to use ~> version constraints only in the root module.

Terraform - Variable defined in "*.auto.tfvars" file, but still cannot be discovered

I have the following directory Structure:
.
├── ./first_terraform.tf
├── ./modules
│   └── ./modules/ssh_keys
│   ├── ./modules/ssh_keys/main.tf
│   ├── ./modules/ssh_keys/outputs.tf
│   └── ./modules/ssh_keys/variables.tf
├── ./terraform.auto.tfvars
├── ./variables.tf
I am trying to pass a variable ssh_key to my child module defined as main.tf inside ./modules/ssh_keys/main.tf
resource "aws_key_pair" "id_rsa_ec2" {
key_name = "id_rsa_ec2"
public_key = file(var.ssh_key)
}
I also have this variable defined both at root and child level variables.tf file. For the value, I have set it in terraform.auto.tfvars as below
# SSH Key
ssh_key = "~/.ssh/haha_key.pub"
I also have a variable defined in root level and child level variables.tf file:
variable "ssh_key" {
type = string
description = "ssh key for EC2 login and checks"
}
My root terraform configuration has this module declared as:
module "ssh_keys" {
source = "./modules/ssh_keys"
}
I first did a terraform init -upgrade on my root level. Then ran terraform refresh and got hit by the following error.
Error: Missing required argument
on first_terraform.tf line 69, in module "ssh_keys":
69: module "ssh_keys" {
The argument "ssh_key" is required, but no definition was found.
Just for reference, line 69 in my root level configuration is where the module declaration has been made. I don't know what I have done wrong here. It seems I have all the variables declared, so am I missing some relationship between root/child module variable passing etc.?
Any help is appreciated! Thanks
I Think I know what I did wrong.
Terraform Modules - as per the documentation requires parents to pass on variables as part of invocation. For example:
module "foo" {
source = "./modules/foo"
var1 = value
var2 = value
}
The above var1, var2 can come from either auto.tfvars file, environment variables (recommended) or even command line -var-file calls. In fact, this is what Terraform calls "Calling a Child Module" here
Once I did that, everything worked like a charm! I hope I did find the correct way of doing things.

How to indicate custom configuration files for terragrunt modules?

I am trying to build Terragrunt script for deploying the infrastructure to Microsoft Azure cloud. Things are working fairly well but I am not able to figure out one thing.
The structure of setup looks something like this:
rootdir
terragrunt.hcl
someconfig.hcl
module1dir
terragrunt.hcl
config.auto.tfvars.json
module2dir
terragrunt.hcl
config.auto.tfvars.json
module3dir
terragrunt.hcl
config.auto.tfvars.json
Each module is configured using Terraform autoload tfvars feature with config.auto.tfvars.json. What I would like is to have these files outside of the directory structure and somehow instruct Terragrunt to apply correct external configuration file to correct submodule.
Any ideas?
I solved this in the following manner:
Define environment variable you plan on using which should contain location to the configuration files. Make sure it is not clashing with anything existing. In this example we will use TGR_CFGDIR. In the external configuration module place the module configuration files and make sure they are properly named. Each file should be named as the module and end with .auto.tfvars.json. So if your module is named foo you should have config file foo.auto.tfvars.json. Change your terragrunt modules (terragrunt.hcl) to have these statements:
locals {
moduleconfig = get_env("TGR_CFGDIR")
modulename = basename(get_terragrunt_dir())
}
generate "configuration" {
path = "config.auto.tfvars.json"
if_exists = "overwrite"
disable_signature = true
contents = file("${local.moduleconfig}/${local.modulename}.auto.tfvars.json")
}
And finally call terragrunt cli like this:
TGR_CFGDIR="<configdir>" terragrunt "<somecommand>"

Terraform modules: correct references of variables?

I'm writing a terraform script to create an EKS cluster with its worker nodes on AWS. First time doing it so I'm a bit confused.
Here is the folder organisation:
├─── Int AWS Account
│ ├─── variables.tf
│ ├─── eks-cluster.tf (refers the modules)
│ ├─── others
│
├─── Prod AWS Account
│ ├─── (will be the same than Int with different settings in variables)
│
├─── ReadMe.md
│
├─── data sources
│
├─── Modules
│ ├─── cluster.tf
│ ├─── worker-nodes.tf
│ ├─── worker-nodes-sg.tf
I am a bit confused regarding how to use and pass variables. Right now, what I'm doing is that I refer to ${var.name} in the module folder, in the eks-cluster.tf, I either put a direct value name = blabla (mostly avoiding it), or refer to the variable again and have a variable file in the account folder.
Is that correct?
I'm not sure if I get your question correctly but in general you would want to keep your module files with variables only, as modules are intended to be generic so you can easily include them in different environments.
When including the module in eks_cluster_int.tf or eks_cluster_prod.tf you would then pass the values for all variables defined in the module itself. This way you can use the environment specific values in the same module.
module "cluster" {
source = "..."
var1 = value1 # directly passing value
var2 = ${var.int_specific_var} # can be defined in variables.tf of environment
...
}
Does this answer your question?

Resources