search

aleksanderaleksic / tgmigrate

7 stars 1 forks source link

readme

tgmigrate

tgmigrate helps you with migrating state in and across terraform state files in a terragrunt environment.

Why?

Migrating state is necessary when refactoring your terragrunt project, and moving resources between state files or within them can be tricky. There are projects like tfmigrate that's targeted towards terraform, but we found it hard to use with terragrunt. Usually we write shell scripts to move or remove state resources with the terraform state commands, but there is no way to store the history to check if the migration have been applied.

Features

Install

Download the latest package from releases and put it in the executable path.

Usage

You first have to set up a .tgmigrate.hcl file:

Example config file:

migration {
  migration = "./migrations"

  history {
    storage "s3" {
      bucket = "airthings-terraform-states-${ACCOUNT}"
      region = "us-east-1"
      assume_role = "${ASSUME_ROLE}"
      key = "history.json"
    }
  }

  state "s3" {
    bucket = "airthings-terraform-states-${ACCOUNT}"
    region = "us-east-1"
    assume_role = "${ASSUME_ROLE}"
  }
}

this file can be specified by using the -c(config) flag,
if not specified tgmigrate will look in the parent folders up to $HOME

migration = "./migration" refers to the directory where the migration files are located, this path is relative to the config file.
history is where you can configure where to store the history, currently its only storage "s3" that's supported.
state is where you configure where your state is located, also here its only state "s3" that's supported.

A nice feature in the config file is the support for variables using the common ${variable_name} syntax, How to populate the variables is described further down.

Next you need to make a migration file:

Under the migration directory you specified in the config file, create a new file and name it something like this: V1__descriptive_text_of_the_migration.hcl.
The V1__ prefix is required and inspired by flyway's versioning concepts, and makes sure that the order of the migrations are executed correctly.

Example migration file:

migration {
  environments = [
    "dev",
    "prod",
    "developer"
  ]
  description = <<EOF
    - Move the rest2 lambda to a separate module.
EOF
}

migrate "move" "rest_2" {
  from {
    state = "us-east-1/apis/rest"
    resource = "aws_lambda_function.rest_2_api_lambda"
  }

  to {
    state = "us-east-1/apis/rest_2"
    resource = "aws_lambda_function.rest_api_lambda"
  }
}

migrate "remove" "api_gateway_integration_for_rest_2" {
  state = "us-east-1/apis/rest"
  resource = "aws_apigatewayv2_integration.rest_2_api"
}

The migration block contains metadata for the migration file. Here you can specify the environments the migration file is intended for, this is optional to include. You can also include a description for the migration but that is optional.

The migrate blocks have 2 supported types: move and remove.
On move you need to specify a from and to block, while on remove you can provide the state and resource variable.
In both cases state refers to the location within the s3_bucket. resource refers to the resource type + name in the state file.

Integrate with terragrunt:

tgmigrate can be integrated with terragrunt's before hook.
There is currently a issue with the integration because there is no option to only run tgmigrate once, this makes terragrunt run tgmigrate for every module.
You can work around this issue by running it manually before running terragrunt apply or you can use the hook on a module that is always running first by using dependencies.

before_hook "plan_migrations" {
  commands = [
    "plan"
  ]
  execute = [
    "tgmigrate",
    "-y",
    "--cv=ACCOUNT_ID=${local.account_id};ASSUME_ROLE=${local.terraform_role_arn}",
    "plan",
    "prod"
  ]
  run_on_error = false
}
before_hook "run_migrations" {
  commands = [
    "apply"
  ]
  execute = [
    "tgmigrate",
    "-y",
    "--cv=ACCOUNT_ID=${local.account_id};ASSUME_ROLE=${local.terraform_role_arn}",
    "apply",
    "prod"
  ]
  run_on_error = false
}

Notice the --cv(config-variables) flag, here we specify the ACCOUNT_ID and the ASSUME_ROLE variable that we use in the config file.
Also notice the prod sub-command, this is telling tgmigrate to only apply migrations for the prod environment.
This means that the migration file above would be applied in this case, because it has prod in the environments list.

Example output

2021/02/01 14:32:16 Downloading us-east-1/apis/rest/terraform.tfstate to /home/user/terragrunt/.tgmigrate_cache/state/us-east-1/apis/rest/terraform.tfstate
2021/02/01 14:32:16 Downloading us-east-1/apis/rest_2/terraform.tfstate to /home/user/terragrunt/.tgmigrate_cache/state/us-east-1/apis/rest_2/terraform.tfstate
Migrations for sample_migration_file.hcl will be applied
Moving aws_lambda_function.rest_2_api_lambda from us-east-1/apis/rest to us-east-1/apis/rest_2 aws_lambda_function.rest_api_lambda
Removing aws_apigatewayv2_integration.rest_2_api from us-east-1/apis/rest