course: Mastering Production Data Streaming Systems with Apache Kafka®

Build an Event Streaming Platform with GitOps (Exercise)

Gilles Philippart

Software Practice Lead

Instructions

We're going to build a pipeline to create both a staging and a production Confluent Cloud data streaming platform.

In a true GitOps fashion, instead of using the Web UI or the CLI to do so, we're going to use Terraform and GitHub actions to build a CI/CD pipeline and automate the deployment.

You will need a Confluent Cloud account, a GitHub account, and an AWS account to store the Terraform state (though you can adapt the terraform code to use a different backend). Make sure to use the promo code at the bottom of this page, it will give you enough credit to do this exercise!

First, create a directory for this exercise.

mkdir data-streaming-platform-pipeline && cd data-streaming-platform-pipeline

We need a Confluent Cloud API key which will be used by Terraform to create and update all Confluent Cloud resources programmatically.

We're going to create this API key via the Confluent Cloud CLI, you can install it with brew install confluentinc/tap/cli or follow the instructions here.

If you prefer, you can also use the Confluent Cloud UI and head over to "Cloud API Keys" in the Administration section. Just make sure to select "Granular access" to create the service account.

Let's create the service account with the CLI:

confluent iam service-account create terraform_runner --description "Service Account used by Terraform to create and update Confluent Cloud resources"

This is the output you should get:

+-------------+--------------------------------+
| ID          | sa-56q6kz                      |
| Name        | terraform_runner               |
| Description | Service Account used by        |
|             | Terraform to create and update |
|             | Confluent Cloud resources      |
+-------------+--------------------------------+

We also must grant this account the OrganizationAdmin role, just use the service account id you've just created above in the command below:

confluent iam rbac role-binding create --principal User:sa-56q6kz --role OrganizationAdmin

Let's create an API key, again, use the service account id you've created:

confluent api-key --service-account sa-56q6kz create --resource cloud --description "Confluent Cloud API key for terraform_runner service account"

You should get something along those lines (note that I've deleted that key long before publishing this exercise).

It may take a couple of minutes for the API key to be ready.
Save the API key and secret. The secret is not retrievable later.
+------------+------------------------------------------------------------------+
| API Key    | DIK7CITBPM5L6KC6                                                 |
| API Secret | RV+GNsKQT6xLjI82wo1ItakGzFeTNwYUzBVKaWvOME+Hh4vdO2ZZp98r6HDzkfmh |
+------------+------------------------------------------------------------------+

Great, we have an API key that we can now use to run Terraform from within the CI/CD pipeline. Keep it in a safe place, we're going to use in the next steps.

Now, create a subdirectory called staging

mkdir staging && cd staging

In this directory, create a main.tf file with the following content:

# Configure the Confluent Provider
terraform {
  required_providers {
    confluent = {
      source  = "confluentinc/confluent"
      version = "1.47.0"
    }
  }
}

# Configure the terraform backend to store the terraform state
terraform {
  backend "s3" {
    bucket = "platform-engineering-terraform-state"
    key    = "terraform/all-state/data-streaming-platform.tfstate"
    region = "us-east-1"
    encrypt = true
  }
}

variable "confluent_cloud_api_key" {
  description = "Confluent Cloud API Key (also referred as Cloud API ID)"
  type        = string
  sensitive = "true"
}

variable "confluent_cloud_api_secret" {
  description = "Confluent Cloud API Secret"
  type        = string
  sensitive = "true"
}

variable "confluent_cloud_environment_name" {
  default = "unnamed"
}

provider "confluent" {
  cloud_api_key       = var.confluent_cloud_api_key
  cloud_api_secret    = var.confluent_cloud_api_secret
}

resource "confluent_environment" "env" {
  display_name = var.confluent_cloud_environment_name
}

resource "confluent_kafka_cluster" "standard" {
  display_name = "standard_kafka_cluster"
  availability = "SINGLE_ZONE"
  cloud        = "AWS"
  region       = "us-east-2"
  standard {}

  environment {
    id = confluent_environment.env.id
  }
}

# We use a dedicated service account, called 'platform-manager', per environment
resource "confluent_service_account" "platform-manager" {
  display_name = "platform-manager-${confluent_environment.env.display_name}"
  description  = "Service account to manage the platform"
}

# The 'platform-manager' account is a cloud cluster admin.
resource "confluent_role_binding" "platform-manager-kafka-cluster-admin" {
  principal   = "User:${confluent_service_account.platform-manager.id}"
  role_name   = "CloudClusterAdmin"
  crn_pattern = confluent_kafka_cluster.standard.rbac_crn
}

resource "confluent_api_key" "platform-manager-kafka-api-key" {
  display_name = "platform-manager-kafka-api-key"
  description  = "Kafka API Key that is owned by 'platform-manager' service account"
  owner {
    id          = confluent_service_account.platform-manager.id
    api_version = confluent_service_account.platform-manager.api_version
    kind        = confluent_service_account.platform-manager.kind
  }

  managed_resource {
    id          = confluent_kafka_cluster.standard.id
    api_version = confluent_kafka_cluster.standard.api_version
    kind        = confluent_kafka_cluster.standard.kind

    environment {
      id = confluent_environment.env.id
    }
  }

  # The goal is to ensure that confluent_role_binding.app-manager-kafka-cluster-admin is created before
  # confluent_api_key.app-manager-kafka-api-key is used to create instances of
  # confluent_kafka_topic, confluent_kafka_acl resources.
  # 'depends_on' meta-argument is specified to avoid having multiple copies of this definition in the configuration
  # which would happen if we specify it in confluent_kafka_topic, confluent_kafka_acl resources instead.
  depends_on = [
    confluent_role_binding.platform-manager-kafka-cluster-admin
  ]
}

If we wanted to, we could install Terraform locally and run the Terraform code from there, but that's exactly what we want to avoid when doing GitOps. Instead, we want a visible, repeatable, audited mechanism to deploy our data streaming platform. It's time for automation!

First, create a .github/workflows directory in your repository. shell cd .. && mkdir -p .github/workflows && cd .github/workflows In this .github/workflows directory, we're going to create 3 GitHub Actions workflow files: ci.yml, cd.yml and promote.yaml

The ci.yml contains the CI workflow which will be run when you open or update a pull request. It runs a lint job and prints the terraform plan in a PR comment:

name: CI

on:
  pull_request:
    types:
      - opened
      - synchronize
    paths-ignore:
      - '**/README.md'

env:
  # We're using AWS S3 as the Terraform backend
  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
  TF_BUCKET_STATE: ${{ vars.TF_BUCKET_STATE }}

  # Credentials for Confluent Cloud
  TF_VAR_CONFLUENT_CLOUD_API_KEY: ${{ secrets.TF_VAR_CONFLUENT_CLOUD_API_KEY }}
  TF_VAR_CONFLUENT_CLOUD_API_SECRET: ${{ secrets.TF_VAR_CONFLUENT_CLOUD_API_SECRET }}

  # Tell Terraform it's running in CI/CD
  TF_IN_AUTOMATION: true

jobs:
  lint:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
        name: Checkout source code

      - uses: actions/cache@v3
        name: Cache plugin dir
        with:
          path: ~/.tflint.d/plugins
          key: tflint-${{ hashFiles('.tflint.hcl') }}

      - uses: terraform-linters/setup-tflint@v3
        name: Setup TFLint
        with:
          tflint_version: v0.44.1

      - name: Show version
        run: tflint --version

      - name: Init TFLint
        run: tflint --init
        env:
          GITHUB_TOKEN: ${{ github.token }}

      - name: Run TFLint
        run: tflint -f compact

  terraform_plan:
    needs: [lint]
    name: "Terraform Plan"
    runs-on: ubuntu-latest
    defaults:
      run:
        shell: bash
        working-directory: ./staging
    steps:
      - name: Checkout the repository to the runner
        uses: actions/checkout@v3

      - name: Setup Terraform with specified version on the runner
        uses: hashicorp/setup-terraform@v2
        with:
          terraform_version: 1.3.0

      - name: Terraform init
        id: init
        run: terraform init -backend-config="bucket=$TF_BUCKET_STATE"

      - name: Terraform validate
        id: validate
        run: terraform validate

      - name: Terraform plan
        id: plan
        run: terraform plan -no-color -input=false -var="confluent_cloud_api_key=$TF_VAR_CONFLUENT_CLOUD_API_KEY" -var="confluent_cloud_api_secret=$TF_VAR_CONFLUENT_CLOUD_API_SECRET"
        continue-on-error: true

      - uses: actions/github-script@v6
        env:
          PLAN: "terraform\n${{ steps.plan.outputs.stdout }}"
        with:
          script: |
            const output = `#### Terraform Format and Style 🖌
            #### Environment ${{ matrix.environment }} (CI)
            #### Terraform Initialization ⚙️\`${{ steps.init.outcome }}\`
            #### Terraform Validation 🤖\`${{ steps.validate.outcome }}\`
            #### Terraform Plan 📖\`${{ steps.plan.outcome }}\`

            <details><summary>Show Plan</summary>

            \`\`\`\n
            ${process.env.PLAN}
            \`\`\`

            </details>
            *Pushed by: @${{ github.actor }}, Action: \`${{ github.event_name }}\`*`;

            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body: output
            })

      - name: Terraform Plan Status
        if: steps.plan.outcome == 'failure'
        run: exit 1

Next, copy the following GitHub Action workflow in a cd.yml file. This file contains the CD workflow which will run when you merge a pull request. It applies the terraform plan to all environments.

name: CD

on:
  pull_request:
    types:
      - closed
    paths-ignore:
      - '**/README.md'

env:
  # We're using AWS S3 as the Terraform backend
  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
  TF_BUCKET_STATE: ${{ vars.TF_BUCKET_STATE }}

  # Credentials for Confluent Cloud
  TF_VAR_CONFLUENT_CLOUD_API_KEY: ${{ secrets.TF_VAR_CONFLUENT_CLOUD_API_KEY }}
  TF_VAR_CONFLUENT_CLOUD_API_SECRET: ${{ secrets.TF_VAR_CONFLUENT_CLOUD_API_SECRET }}

  # Tell Terraform it's running in CI/CD
  TF_IN_AUTOMATION: true

jobs:
  deploy_streaming_platform:
    name: "Deploy ${{ matrix.environment }} platform"
    if: github.event.pull_request.merged
    strategy:
      max-parallel: 1
      matrix:
        environment: [ staging, prod ]
    runs-on: ubuntu-latest
    steps:
      - name: Checkout the repository to the runner
        uses: actions/checkout@v3

      - name: Check terraform files existence
        id: check_files
        uses: andstor/file-existence-action@v2
        with:
          files: "${{ matrix.environment }}/*.tf"
          fail: true

      - name: Setup Terraform with specified version on the runner
        uses: hashicorp/setup-terraform@v2
        with:
          terraform_version: 1.5.4

      - name: Terraform init
        id: init
        run: terraform init -input=false -backend-config="bucket=$TF_BUCKET_STATE"
        working-directory: ${{ matrix.environment }}

      - name: Terraform validate
        id: validate
        run: terraform validate
        working-directory: ${{ matrix.environment }}

      - name: Terraform workspace
        id: workspace
        run: terraform workspace select -or-create ${{ matrix.environment }}
        working-directory: ${{ matrix.environment }}

      - name: Terraform plan
        id: plan
        run: terraform plan -no-color -input=false
          -var="confluent_cloud_api_key=$TF_VAR_CONFLUENT_CLOUD_API_KEY"
          -var="confluent_cloud_api_secret=$TF_VAR_CONFLUENT_CLOUD_API_SECRET"
          -var="confluent_cloud_environment_name=${{ matrix.environment }}"
        continue-on-error: true
        working-directory: ${{ matrix.environment }}

      - uses: actions/github-script@v6
        env:
          PLAN: "terraform\n${{ steps.plan.outputs.stdout }}"
        with:
          script: |
            const output = `#### Terraform Format and Style 🖌
            #### Environment ${{ matrix.environment }} (CD)
            #### Terraform Initialization ⚙️\`${{ steps.init.outcome }}\`
            #### Terraform Validation 🤖\`${{ steps.validate.outcome }}\`
            #### Terraform Plan 📖\`${{ steps.plan.outcome }}\`

            <details><summary>Show Plan</summary>

            \`\`\`\n
            ${process.env.PLAN}
            \`\`\`

            </details>
            *Pushed by: @${{ github.actor }}, Action: \`${{ github.event_name }}\`*`;

            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body: output
            })

      - name: Terraform Apply
        if: github.ref == 'main' && github.event.pull_request.merged
        run: terraform apply -auto-approve -input=false
          -var="confluent_cloud_api_key=$TF_VAR_CONFLUENT_CLOUD_API_KEY"
          -var="confluent_cloud_api_secret=$TF_VAR_CONFLUENT_CLOUD_API_SECRET"
          -var="confluent_cloud_environment_name=${{ matrix.environment }}"
        working-directory: ${{ matrix.environment }}

Finally, here's the promote.yml file below. This workflow copies the content of the staging directory to the prod directory and runs whenever code is pushed to the main branch. We assume that you'll only make pull requests with changes in the staging directory or in the prod/specific directory.

name: Promote staging to prod
run-name: ${{ github.actor }} wants to promote Staging to Prod 🚀
on:
  push:
    branches:
      - main
    paths-ignore:
      - '**/README.md'

jobs:
  promote:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v3

      - run: echo "🔎 Compare and sync directories"
      - name: Compare and sync directories
        run: |
          # Exclude the 'specific' directory and compare staging and prod
          DIFF=$(rsync --stats --recursive --dry-run --exclude=specific ./staging/ ./prod/ | awk '/files transferred/{print $NF}')

          # If DIFF is not 0, means there are differences
          if [ "$DIFF" -ne 0 ]; then
            # If there are differences, use rsync to copy files excluding the 'specific' directory
            rsync --checksum --recursive --exclude=specific ./staging/ ./prod/
          fi

      - name: Set current date as env variable
        run: echo "NOW=$(date +'%Y-%m-%d-%H%M%S')" >> $GITHUB_ENV

      - name: Create Pull Request
        uses: peter-evans/create-pull-request@v5
        with:
          token: ${{ secrets.GITHUB_TOKEN }}
          commit-message: Sync staging to prod
          title: "Sync staging to prod ${{env.NOW}}"
          body: This PR is to sync the staging directory to the prod directory.
          branch: "sync-staging-to-prod-${{env.NOW}}"

We also need to create a specific subdirectory in each environment directory if we need environment specific code which will escape the copying from staging to prod done by the promote action.

cd ../..
mkdir -p staging/specific && touch staging/specific/.gitkeep 
mkdir -p prod/specific && touch prod/specific/.gitkeep

The last file we need is an AWS policy to grant read/write access to the S3 bucket which will store the Terraform State. We'll use it later when we do the initial setup.

Create the following s3_bucket_full_access_policy.json at the top of your repository directory:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": "s3:*",
      "Resource": [
        "arn:aws:s3:::platform-engineering-terraform-state",
        "arn:aws:s3:::platform-engineering-terraform-state/*"
      ]
    }
  ]
}

Now, commit and push your changes (change your_user with your own username)

git init
git add .github staging prod *.json

If you run git status this is what you should get:

On branch main

No commits yet

Changes to be committed:
  (use "git rm --cached <file>..." to unstage)
	new file:   .github/workflows/cd.yml
	new file:   .github/workflows/ci.yml
	new file:   .github/workflows/promote.yml
	new file:   prod/specific/.gitkeep
	new file:   s3_bucket_full_access_policy.json
	new file:   staging/main.tf
	new file:   staging/specific/.gitkeep

Before committing and pushing those changes, we must allow GitHub Actions to create pull requests. Under the "Actions -> General -> Workflow Permissions" settings of your freshly created GitHub repository:

enable the "Allow GitHub Actions to create and approve pull requests" option
allow GITHUB_TOKEN to read and write to your repository.
click the Save button.

github-actions-settings

With these two options enabled, our promote workflow will be able to create pull requests.

Run the following aws CLI commands to create the S3 bucket to store the state:

aws s3api create-bucket --bucket "platform-engineering-terraform-state"

aws iam create-user --user-name "terraform-user"

aws iam create-policy --policy-name S3FullAccessPolicy --policy-document file://s3_bucket_full_access_policy.json
# arn:aws:iam::321751332100:policy/S3FullAccessPolicy

aws iam attach-user-policy --policy-arn arn:aws:iam::321751332100:policy/S3FullAccessPolicy --user-name terraform-user

Add the following entries in your GitHub repo settings, under "Secrets and Variables > Actions":

Add a repository variable called TF_BUCKET_STATE set to platform-engineering-terraform-state.
Add repository secrets AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY set your AWS account API key (use a dedicated IAM user, not your root account)
Add repository secrets TF_VAR_CONFLUENT_CLOUD_API_KEY and TF_VAR_CONFLUENT_CLOUD_API_SECRET set to the API credentials created earlier for your terraform-user.

Now, let's commit and push this code to GitHub.

git checkout -b automation
git commit -m "Automate the platform creation"
git remote add origin git@github.com:your_user/data-streaming-platform-pipeline.git
git push -u origin automation

Once you've pushed, open a Pull Request for the automation branch. Doing so will trigger the CI workflow.

ci-workflow

The CI workflow lints the Terraform code and gives you a preview of the plan directly in the PR as a comment:

ci-workflow-terraform-plan

When you merge this Pull Request, the promote workflow will trigger and create another Pull Request which copies the whole content of the staging directory to the prod directory, except any files sitting in the specific subdirectory.

promote-workflow

If you open this Pull Request, you can see that the proposed change is the addition of the main.tf file under the prod directory, as expected.

promote-pr

As soon as you'll merge that promotion Pull Request, the CD workflow will trigger a job to Terraform all environments:

cd-workflow-outcome

Now, if you head over to Confluent Cloud, you can see that both environments have been created and have the same exact configuration.

Congratulations!

confluent-cloud-envs

Do you have questions or comments? Join us in the #confluent-developer community Slack channel to engage in discussions with the creators of this content.

Use the promo codes KAFKAPROD101 & CONFLUENTDEV1 to get $25 of free Confluent Cloud usage and skip credit card entry.

Get Started

Be the first to get updates and new content

We will only share developer content and updates, including notifications when new content is added. We will never send you sales emails. 🙂 By subscribing, you understand we will process your personal information in accordance with our Privacy Statement.

NEWApache Flink® SQL

NEWApache Flink® Table API: Processing Data Streams in Java

NEWDesigning Event-Driven Microservices

NEWApache Flink® 101

NEWBuilding Flink® Apps in Java

NEWMastering Production Data Streaming Systems with Apache Kafka®

Kafka® 101

Kafka® Connect 101

Kafka Streams 101

Schema Registry 101

ksqlDB 101

Data Mesh 101

NEWApache Flink® SQL

NEWApache Flink® Table API: Processing Data Streams in Java

NEWDesigning Event-Driven Microservices

NEWApache Flink® 101

NEWBuilding Flink® Apps in Java

NEWMastering Production Data Streaming Systems with Apache Kafka®

Kafka® 101

Kafka® Connect 101

Kafka Streams 101

Schema Registry 101

ksqlDB 101

Data Mesh 101

Articles

Patterns

FAQs

Blog

NEWStreamables

NEWLearn More

Articles

Patterns

FAQs

Blog

NEWStreamables

NEWLearn More

Language Guides

Tutorials

Demos

Language Guides

Tutorials

Demos

Meetups

Ask the Community

Community Catalysts

NEWCommunity Use Cases

Confluent Developer Newsletter

Data Streaming Awards

NEWCurrent 2024

Kafka Summit 2024 - Bangalore

Kafka Summit 2024 - London

Current 2023

Kafka Summit 2023

Meetups

Ask the Community

Community Catalysts

NEWCommunity Use Cases

Confluent Developer Newsletter

Data Streaming Awards

NEWCurrent 2024

Kafka Summit 2024 - Bangalore

Kafka Summit 2024 - London

Current 2023

Kafka Summit 2023

NEWApache Flink® SQL

NEWApache Flink® Table API: Processing Data Streams in Java

NEWDesigning Event-Driven Microservices

NEWApache Flink® 101

NEWBuilding Flink® Apps in Java

NEWMastering Production Data Streaming Systems with Apache Kafka®

Kafka® 101

Kafka® Connect 101

Kafka Streams 101

Schema Registry 101

ksqlDB 101

Data Mesh 101

Articles

Patterns

FAQs

Blog

Modules: Start from lesson 1
Total 10