Terraform template recipes

DETAILS: Tier: Free, Premium, Ultimate Offering: SaaS, self-managed

WARNING: The Terraform CI/CD templates are deprecated and will be removed in GitLab 18.0. See the deprecation announcement for more information.

You can customize your Terraform integration by adding the recipes on this page to your pipeline.

If you'd like to share your own Terraform configuration, consider contributing a recipe to this page.

Enable a `terraform destroy` job

Add the following snippet to your .gitlab-ci.yml:

include:
  - template: Terraform.latest.gitlab-ci.yml

destroy:
  extends: .terraform:destroy

The destroy job is part of the cleanup stage. Like the deploy job, the destroy job is always manual and is not tied to the default branch.

To connect the destroy job to the GitLab environment:

include:
  - template: Terraform.latest.gitlab-ci.yml

deploy:
  environment:
    name: $TF_STATE_NAME
    action: start
    on_stop: destroy

destroy:
  extends: .terraform:destroy
  environment:
    name: $TF_STATE_NAME
    action: stop

In this configuration, the destroy job is always created. However, you might want to create a destroy job only if certain conditions are met.

The following configuration creates a destroy job, runs a destroy plan and omits the deploy job only if TF_DESTROY is true:

include:
  - template: Terraform.latest.gitlab-ci.yml

build:
  rules:
    - if: $TF_DESTROY == "true"
      variables:
        TF_CLI_ARGS_plan: "-destroy"
    - when: on_success

deploy:
  environment:
    name: $TF_STATE_NAME
    action: start
    on_stop: destroy
  rules:
    - if: $TF_DESTROY == "true"
      when: never
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH && $TF_AUTO_DEPLOY == "true"
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
      when: manual

destroy:
  extends: .terraform:destroy
  dependencies:
    - build
  variables:
    TF_CLI_ARGS_destroy: "${TF_PLAN_CACHE}"
  environment:
    name: $TF_STATE_NAME
    action: stop
  rules:
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH && $TF_DESTROY == "true"
      when: manual

This configuration has a known issue: when the destroy job is not in the same pipeline as the deploy job, the on_stop environment action does not work.

Run a custom `terraform` command in a job

To define a job that runs a custom terraform command, the gitlab-terraform wrapper can be used in any job:

include:
  - template: Terraform.latest.gitlab-ci.yml

state-list:
  stage: validate # you can use any stage, just make sure to define it
  script: gitlab-terraform state list

The gitlab-terraform command sets up a terraform command and runs it with the given arguments.

To run this job in the Terraform state-specific resource group, assign the job with resource_group:

include:
  - template: Terraform.latest.gitlab-ci.yml

state-list:
  stage: validate # you can use any stage, just make sure to define it
  resource_group: ${TF_STATE_NAME}
  script: gitlab-terraform state list

Add custom debug tools to jobs

The default image used by Terraform template jobs contains only minimal tooling. However, you might want to add additional tools for debugging.

To add an additional tool:

Install the tool in the before_script of a job or pipeline.
Use the tool in the script or after_script block.
- If you use the script block, be sure to re-add the template job commands.

For example, the following snippet installs bash and jq in the before_script for all jobs in the pipeline:

include:
  - template: Terraform.latest.gitlab-ci.yml

default:
  before_script: apk add --update bash jq

To add it to only the build and deploy jobs, add it to those jobs directly:

include:
  - template: Terraform.latest.gitlab-ci.yml

build:
  before_script: apk add --update bash jq

deploy:
  before_script: apk add --update bash jq

Add custom container images

For debug tools and simple installations, you should add a custom debug tool to your job. If your tool is complex or benefits from caching, you can create a custom container image based on the gitlab-terraform images. You can use your custom image in subsequent Terraform jobs.

To define a custom container image:

Define a new Dockerfile with custom tooling. For example, install bash and jq in .gitlab/ci/Dockerfile:

FROM registry.gitlab.com/gitlab-org/terraform-images/stable:latest

RUN apk add --update bash jq

In a new job, define a prepare stage that builds the image whenever the Dockerfile changes.
- The built image is pushed to the GitLab container registry. A tag is applied to indicate whether the image was built from a merge request or from the default branch.
Use your image in your Terraform jobs, such as build and deploy.
- You can combine your image with specialized before_script configurations to perform setup commands, like to generate inputs for Terraform.

For example, a fully functioning pipeline configuration might look like:

include:
  - template: Terraform.latest.gitlab-ci.yml

variables:
  IMAGE_TAG: latest

workflow:
  rules:
    - if: $CI_MERGE_REQUEST_IID
      changes:
        - .gitlab/ci/Dockerfile
      variables:
        IMAGE_TAG: ${CI_COMMIT_REF_SLUG}
    - when: always

stages:
  - prepare
  - validate
  - test
  - build
  - deploy
  - cleanup

prepare:image:
  needs: []
  stage: prepare
  image:
    name: gcr.io/kaniko-project/executor:v1.9.0-debug
    entrypoint: [""]
  rules:
    # Tag with the commit SHA if we're in an MR
    - if: $CI_MERGE_REQUEST_IID
      changes:
        - .gitlab/ci/Dockerfile
      variables:
        DOCKER_TAG: $CI_COMMIT_REF_SLUG
    # If we're on our main branch, tag with "latest"
    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
      changes:
        - .gitlab/ci/Dockerfile
      variables:
        DOCKER_TAG: latest
  before_script:
    # Authenticate to the docker registry and dependency proxy
    - echo "{\"auths\":{\"$CI_REGISTRY\":{\"auth\":\"$(printf "%s:%s" "${CI_REGISTRY_USER}" "${CI_REGISTRY_PASSWORD}" | base64 | tr -d '\n')\"}}}" > /kaniko/.docker/config.json
  script:
    - /kaniko/executor
      --context "${CI_PROJECT_DIR}/.gitlab/ci"
      --cache=true
      --dockerfile "${CI_PROJECT_DIR}/.gitlab/ci/Dockerfile"
      --destination "${CI_REGISTRY_IMAGE}:${DOCKER_TAG}"

build:
  image: ${CI_REGISTRY_IMAGE}:${IMAGE_TAG}

deploy:
  image: ${CI_REGISTRY_IMAGE}:${IMAGE_TAG}

For an example repository, see the GitLab Terraform template usage project.

Automatically deploy from the default branch

You can automatically deploy from the default branch by setting the TF_AUTO_DEPLOY variable to "true". All other values are interpreted as "false".

variables:
  TF_AUTO_DEPLOY: "true"

include:
  - template: Terraform.latest.gitlab-ci.yml

Deploy Terraform to multiple environments

You can run pipelines in multiple environments, each with a unique Terraform state.

stages:
  - validate
  - test
  - build
  - deploy

include:
  - template: Terraform/Base.latest.gitlab-ci.yml
  - template: Jobs/SAST-IaC.latest.gitlab-ci.yml

variables:
  # x prevents TF_STATE_NAME from beeing empty for non environment jobs like validate
  # wait for https://gitlab.com/groups/gitlab-org/-/epics/7437 to use variable defaults
  TF_STATE_NAME: x${CI_ENVIRONMENT_NAME}
  TF_CLI_ARGS_plan: "-var-file=vars/${CI_ENVIRONMENT_NAME}.tfvars"

fmt:
  extends: .terraform:fmt
validate:
  extends: .terraform:validate

plan dev:
  extends: .terraform:build
  environment:
    name: dev
plan prod:
  extends: .terraform:build
  environment:
    name: prod

apply dev:
  extends: .terraform:deploy
  environment:
    name: dev

apply prod:
  extends: .terraform:deploy
  environment:
    name: prod

This configuration is modified from the base GitLab template.