How To Have Terragrunt Auto Set "AWS_PROFILE" environment variable?

Question

I cannot for the life of me get this to work. I need to have the AWS_PROFILE environment variable set to get terragrunt to run properly. If I run:

export AWS_PROFILE=myprofile; terragrunt plan

That will work, but that's not what I'm after I want to just run:

terragrunt plan

and have that auto choose the correct aws profile I should be using. Here is what I have:

generate "provider" {
  path = "provider.tf"
  if_exists = "overwrite_terragrunt"
  contents = <<EOF
provider "aws" {
  region  = "${local.region}"
  profile = "${trimspace(run_cmd("bash", "${get_parent_terragrunt_dir()}/../../set_profile.sh",local.profile))}"
}
EOF
}
remote_state {
  backend = "s3"
  generate = {
    path      = "backend.tf"
    if_exists = "overwrite"
  }
  config = {
    ...
    ...
    region         = local.region
    profile        = local.profile
    ...
    ...
  }
}

It always throws the error on me:

Error finding AWS credentials (did you set the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables?): NoCredentialProviders: no valid providers in chain. Deprecated.
        For verbose messaging see aws.Config.CredentialsChainVerboseErrors

The set_profile.sh script is the following:

#!/bin/bash

VALUE=$(echo $1 | sed $'s/\r//')
export AWS_PROFILE=$VALUE
echo "$AWS_PROFILE"

If I echo out my AWS_PROFILE it's still blank. So it's like the run command isn't actually saving the export value to my console.

What am I doing wrong? Has anyone actually successfully been able to dynamically set their AWS_PROFILE with terragrunt?

Answer 1

That's my solution. I have the following structure:

<project>
    |-- <region1>
    |-- <region2>
    |-- account.hcl
terragrunt.hcl

In account.hcl

locals {
  aws_profile_name = "myprofile"
}

In main terragrunt.hcl

locals {
  # Automatically load account-level variables
  account_vars = read_terragrunt_config(find_in_parent_folders("account.hcl"))

  aws_profile = local.account_vars.locals.aws_profile_name
}

terraform {
  extra_arguments "aws_profile" {
    commands = [
      "init",
      "apply",
      "refresh",
      "import",
      "plan",
      "taint",
      "untaint"
    ]

    env_vars = {
      AWS_PROFILE = "${local.aws_profile}"
    }
  }
}

remote_state {
  ...
  config = {
    ...
    profile = "${local.aws_profile}"
  }
}

generate "provider" {
  ...
  contents 
  contents  = <<EOF
provider "aws" {
  profile = "${local.aws_profile}"
}
EOF
}
...

Answer 2

This post helped me figure out my problem:

I had forget the fact that my config have 2 AWS connections to setup

• the backend
• the provider

Hence the AWS profile had to be set twice:

• in the remote_state

  remote_state {
    backend = "s3"
    config = {
      ...
      profile = local.profile
      ...
    }
  }
  

• in the provider.tf

  generate "provider" {
    path      = "provider.tf"
    if_exists = "skip"
    contents  = <<EOF
  provider "aws" {
    ...
    profile = "${local.profile}"
    ...
  }
  EOF
  }
  

Hope this saves someone all the time I've wasted today!