Skip to content

Latest commit

 

History

History
276 lines (196 loc) · 11.2 KB

README.md

File metadata and controls

276 lines (196 loc) · 11.2 KB

Codacy Badge Codacy Badge Build Status License PyPI Python Versions Downloads

Terrawrap

Set of Python-based CLI tools for working with Terraform configurations in bulk

About Amplify

Amplify builds innovative and compelling digital educational products that empower teachers and students across the country. We have a long history as the leading innovator in K-12 education - and have been described as the best tech company in education and the best education company in tech. While others try to shrink the learning experience into the technology, we use technology to expand what is possible in real classrooms with real students and teachers.

Learn more at https://www.amplify.com

Table of Contents

Features

  1. auto.tfvars inheritance. Terrawrap makes it easier to share variables between Terraform directories through inheritance of auto.tfvars files.

  2. Remote backend generation. Terrawrap makes it easier to work with remote state backends by generating configuration for them.

  3. Repository level plan/apply. Terrawrap provides commands for running plan/apply recursively on a entire repository at once.

  4. Repository level dependency visualization. Terrawrap provides commands for displaying the order of applies in human readable output.

  5. Automatically download third-party Terraform plugins

Goals

  1. Make Terraform DRY for large organizations. A Terraform best practices is to break up Terraform configs into many small state files. This leads to an explosion in boilerplate code when using Terraform in large organizations with 100s of state files. Terrawrap reduces some boilerplate code by providing auto.tfvars inheritance and generating backend configurations.

  2. Make Terraform code easier to manage. Terraform only runs commands on a single directory at a time. This makes working with hundreds of terraform directories/state files hard. Terrawrap provides utilities for running commands against an entire repository at once instead of one directory at a time.

  3. All Terraform code should be valid Terraform. Any Terraform code used with Terrawrap should be runnable with Terraform by itself without the wrapper. Terrawrap does not provide any new syntax.

  4. Terrawrap is not a code generator. Generated code is harder to read and understand. Code generators tend to lead to leaky abstractions that can be more trouble than they are worth. However, Terrawrap does generate remote backend configs as a workaround to Terraform's lack of support for variables in backend configs (See hashicorp/terraform#13022). We expect this to be the only instance of code generation in Terrawrap.

Getting Started

Prerequisites

Terrawrap requires Python 3.7.0 or higher to run.

Installing

This package can be installed using pip

pip3 install terrawrap

You should now be able to use the tf command.

Building From Source

For development, tox>=2.9.1 is recommended.

Running Tests

Terrawrap uses tox. You will need to install tox with pip install tox. Running tox will automatically execute the unit tests.

You can also run them individually with the -e argument.

For example, tox -e py37-unit will run the unit tests for python 3.7

To see all the available options, run tox -l.

Configuration

.tf_wrapper

Terrawrap can be configured via a .tf_wrapper file. The wrapper will walk the provided configuration path and look for .tf_wrapper files. The files are merged in the order that they are discovered. Consider the below example:

foo
├── bar
│   └── .tf_wrapper
└── .tf_wrapper

If there are conflicting configurations between those two .tf_wrapper files, the .tf_wrapper file in foo/bar will win.

The following options are supported in .tf_wrapper:

configure_backend: True # If true, automatically configure Terraform backends.
backend_check: True # If true, require this directory to have a terraform backend configured

envvars:
  <NAME_OF_ENVVAR>:
    source: # The source of the envvar. One of `['ssm', 'text', 'unset']`.
    path: # If the source of the envvar is `ssm`, the SSM Parameter Store path to lookup the value of the environment variable from.
    value: # if the source of the envvar is `text`, the string value to set as the environment variable.
    # If the source is unset, any previous value for the environment variable is removed and the environment variable will not be set.

plugins:
    <NAME_OF_PLUGIN>: <plugin url>

Plugins

Terrawrap supports automatically downloading provider plugins by configuring the .tf_wrapper file as specified above. This is a temporary workaround until Terraform 0.13 is released with built-in support for automatically downloading plugins and plugin registries are available for hosting private plugins.

Terrawrap will first try to download platform specific versions of plugins by downloading them from <plugin url>/<system type>/<architecture type>. If Terrawrap is unable to download from the platform specific URL then it will try to download directly from the given plugin url directly instead.

For example, the following config on a Mac

plugins:
    foo: http://example.com/foo

Terrawap will first try to download from http://example.com/foo/Darwin/x86_64. If that request fails then Terrawrap will try http://example.com/foo instead.

Autovars

Terrawrap automatically adds -var-file arguments to any terraform command by scanning for *.auto.tfvars files in the directory structure.

For example, the following command tf config/foo/bar apply with the following directory structure:

config
├── foo
|   └── bar
|   │  ├── baz.tf
|   │  └── bar.auto.tfvars
|   └── foo.auto.tfvars
└── config.auto.tfvars

will generate the following command:

terraform apply -var-file config/config.auto.tfvars \
    -var-file config/foo/foo.auto.tfvars \
    -var-file config/foo/bar/bar.auto.tfvars

Backend Configuration

Terrawrap supports automatically configuring backends by injecting the appropriate -backend-config args when running init

For example, the Terrawrap command tf config/foo/bar init will generate a Terraform command like below if using an AWS S3 remote state backend

terraform init -reconfigure \
    -backend-config=dynamodb_table=<lock table name> \
    -backend-config=encrypt=true \
    -backend-config=key=config/foo/bar.tfstate \
    -backend-config=region=<region name> \
    -backend-config=bucket=<state bucket name> \
    -backend-config=skip_region_validation=true \
    -backend-config=skip_credentials_validation=true

Terrawrap configures the backend by looking for .tf_wrapper files in the directory structure. Either s3 or gcs are supported. See the relevant Terraform documentation for the options available for each type of backend: https://www.terraform.io/docs/backends/types/s3.html#configuration-variables https://www.terraform.io/docs/backends/types/gcs.html#configuration-variables

S3 Backend

backends:
    s3:
        region:
        role_arn:
        bucket:
        dynamodb_table:
Option Name Required Purpose
bucket Yes Name of S3 Bucket
region Yes AWS Region that S3 state bucket and DynamoDB lock table are located in
dynamodb_table No DynamoDB table to use for state locking. Locking is disable if lock_table is not set
role_arn No AWS role to assume when reading/writing to S3 bucket and lock table

The S3 state file key name is generated from the directory name being used to run the terraform command. For example, tf config/foo/bar init uses a state file with the key config/foo/bar.tfstate in S3

GCS Backend

backends:
    gcs:
        bucket:
Option Name Required Purpose
bucket Yes Name of GCS Bucket

Commands

tf

tf <directory> <terraform command> runs a terraform command for a given directory that contains *.tf files. Terrawrap automatically includes autovars as described above when running the given command. Any Terraform command is supported

plan_check

plan_check <directory> runs terraform plan recursively for all child directories starting at the given directory. plan_check uses git to identify which files have changed compared with the master branch. It will then run plan on any directory that contains tf files with the following criteria

  1. A directory that has files that changed
  2. A directory that is symlinked to a directory that has files changed
  3. A directory with symlinked files that are linked to files that changed
  4. A directory that that uses a Terraform module whose source changed
  5. A directory with Terraform files that refer to an autovar file that changed

backend_check

backend_check [directory] verifies that all directories under the given directory that contain .tf files also have Terraform Backends defined.