Enterprise Maturity Workshop: Terraform

avatar-edunham.jpg E. Dunham
20 MIN READ
Enterprise Maturity Workshop: Terraform

This workshop is part of our Enterprise-Ready Workshop series. Follow along to get familiar with managing identity as code using Okta’s Terraform provider, so you can assist enterprise customers and simplify any identity setup steps that your product might require. In this workshop, you’ll use Terraform to manage users and groups in an Okta Organization, while practicing beginner and intermediate Terraform skills.

Posts in the on-demand workshop series
1. How to Get Going with the On-Demand SaaS Apps Workshops
2. Enterprise-Ready Workshop: Authenticate with OpenID Connect
3. Enterprise-Ready Workshop: Manage Users with SCIM
4. Enterprise Maturity Workshop: Terraform
5. Enterprise Maturity Workshop: Automate with no-code Okta Workflows
6. How to Instantly Sign a User Out across All Your App

The steps in this workshop are also demonstrated in the accompanying video, so that you can follow along in whatever medium is the best fit for your current learning needs.

Table of Contents

1: Terraform automates manual work

Terraform is a general-purpose Infrastructure As Code tool. Okta’s Terraform module lets you write code to accomplish tasks that would otherwise require manual work in the Okta Admin Console. Moving administration tasks to Terraform offers several benefits over doing them by hand:

  • Testing and review: Terraform can guarantee that you make the exact same change to production as to your test environment. You can use your existing code review processes to make sure the right people have approved a change when it happens in Terraform, as well.
  • CI/CD Automation: Terraform lets you add Okta tasks to your continuous integration and continuous delivery pipelines. A Terraform integration in your Okta org has limited scopes, just like a human user, so your automation can only make the kinds of changes that you expect it to.
  • Sharing configuration: If you want to help someone else to make a specific change in their Okta organization, you can give them Terraform code which describes that change. Terraform can make the exact same change anywhere it’s run, whereas telling a person how to configure their organization through the admin console increases opportunities for human error.
  • Auditability: If you already have tools which back up and track changes to your codebase, managing Terraform code with the same processes lets you enjoy the same benefits for your infrastructure. Keeping infrastructure code in source control helps you track exactly how your infrastructure has changed, when each change was made, and who did it.

If you’re new to Terraform, start with Hashicorp’s Terraform tutorials. This workshop assumes that you already have Terraform installed on your computer, and that you have a free Okta Developer Account.

2: Key Terraform concepts

Terraform code describes a desired state for cloud infrastructure. After you write your Terraform code, the terraform plan command predicts what changes will be required to make the infrastructure match the code, and the terraform apply command changes the infrastructure.

A Terraform project is a directory containing .tf files. These Terraform files contain providers, resources, and sometimes variables and data sources too. After running terraform init in a Terraform project, Terraform downloads all required providers to the .terraform directory in that project.

Terraform also stores data about the state of the infrastructure by creating the file terraform.tfstate in the project. Every piece of infrastructure should be represented in only one terraform.tfstate file. If several collaborators work on the same Terraform codebase together, it’s best to use Terraform Cloud or a similar service to share a single terraform.tfstate file.

3: Set up the Okta Terraform provider

Before you can use Terraform to manage resources in your Okta organization, you must provide it with appropriate credentials. The following steps will connect a Terraform project on your local system to your Okta Developer account.

3.1: Manage secrets appropriately

Setting up any Terraform provider involves working with some secrets that can grant access to your infrastructure. The Okta terraform provider uses a client ID and private key. Anyone who has these secrets can act on behalf of Terraform in your Okta organization. Secrets are replaced by placeholder values in the code examples that follow.

When deciding how to handle any secret information, consider the potential impact if it was misused. The impact of leaking a credential to a sandbox account where you test your code may be relatively low, whereas mishandling production secrets could cause extensive downtime and permanent reputation loss for your company.

When you work with Terraform, you must manage secrets as carefully as you protect every other means of accessing your infrastructure. If you store your Terraform code in source control, avoid committing secrets. If you’re running Terraform in automation, find an appropriate secrets management solution to handle the sensitive credentials.

3.2: Create Terraform files

Create a new directory to start your Terraform project in. Pick a name that makes sense: okta-terraform-workshop is a safe bet. In the terminal, you can do this with mkdir okta-terraform-workshop; cd okta-terraform-workshop.

Check that Terraform is installed and up to date with terraform -v. If you’re on an old version of Terraform, follow the instructions in the upgrade message to get the latest version.

In your favorite editor or IDE, create and open the file main.tf in your Terraform workshop directory. When you’re experimenting with Terraform, everything can go in main.tf at first. As main.tf gets too large to work with easily, it can be refactored into smaller files, and eventually into separate modules, for ease of maintenance.

First, instruct Terraform to find and use the Okta provider by giving the project a required_providers block at the top of main.tf:

terraform {
  required_providers {
    okta = {
      source = "okta/okta"
    }
  }
}

If you ever need a reminder of how the provider setup code should look, check the first page of the Okta provider docs on the Terraform registry. Whenever you see registry.terraform.io/providers in the URL of a search result, you know you’ve found the docs which the provider’s authors shipped alongside its code.

Specifying Okta as a required provider ensures that Terraform will download the provider once you run terraform init.

Can the Terraform code in this project manage resources in your Okta organization yet? If you wanted to make changes in an Okta organization through the admin console, you would need to know how to use Okta, but you would also need credentials to log in. Similarly, the Terraform project needs credentials for the Okta organization that you want it to manage.

The provider configuration which tells Terraform how to access the desired Okta organization will go in main.tf:

provider "okta" {
  org_name    = ""
  base_url    = "okta.com"
  client_id   = ""
  scopes      = ["okta.groups.manage", "okta.users.manage", "okta.policies.manage"]
  private_key = ""
}

All of those fields are mandatory, and you will fill out the missing values in the next steps. The org_name will come from the URL of your Okta Developer Account, and the client_id and private_key will come from the App Integration. You will create and download a PKCS key and convert it into the RSA key format for Terraform to use.

Scopes describe categories of changes that Terraform is allowed to make with the Okta Provider. This list of scopes will appear twice: Once here in the provider definition, and again in the Okta App Integration. The 3 scopes provided will be enough to do the exercises in this workshop. Later on, you will explore what would happen if the scopes had been set up wrong at this step.

3.3: Create the Okta app integration

In a web browser, log in to your Okta Developer Account at developer.okta.com. If you don’t have an account yet, now is a great time to create one. Developer accounts are a safe place to experiment without the risk of accidentally changing your production infrastructure. Developer accounts don’t have a time limit, so you can always test your code before taking it to production.

After logging in to your Developer Account, note the landing page URL. It looks like https://dev-1234567890-admin.okta.com/admin/getting-started. Everything after https:// and before -admin is your org_name, and can be copied and pasted into your Terraform provider definition in main.tf.

In the sidebar of the admin console, click Applications, and click Applications under that. The Applications page lists which apps you have installed, and lets you create new ones.

Click the blue “Create App Integration” button, and select “API Services” as the application type. For the application’s name, replace “My API Services App” with “Terraform Workshop” or something similarly descriptive. Save the application.

In the General tab of the application, the Client ID is shown. Copy it to the client_id field in your Terraform provider configuration.

Below the client ID, the “client authentication” radio button defaults to “Client secret”. Terraform needs to use “Public key / Private key” authentication, so click the edit button next to the “client credentials” section heading, and switch the authentication to “public key / private key”. Click “add key”, generate a new key, and save the private key file as pkcs.pem in your Terraform project directory. This key must be converted to RSA format for Terraform to use.

In the browser, save your changes to the general settings of the API Service App, and navigate to the app’s Okta API Scopes. Grant okta.groups.manage, okta.policies.manage, and okta.users.manage for this workshop.

In the terminal, cd okta-terraform-workshop and convert the PKCS-1 key to an RSA (PKCS-8) key:

$ openssl rsa -in pkcs.pem -out rsa.pem

3.4: Configure the Okta Terraform provider

After setting up the app integration, you have found all the values required by the provider configuration! In main.tf, check that your provider block contains the org_name, client_id, and private_key file location. Tell it where to find the private key that you converted. With all the values filled out, the provider configuration looks like this:

provider "okta" {
  org_name    = "dev-1234567890"
  base_url    = "okta.com"
  client_id   = "00abc123FIXME00"
  scopes      = ["okta.groups.manage", "okta.users.manage", "okta.policies.manage"]
  private_key = "${path.module}/rsa.pem"
}

4: Manage users and groups

With the provider configured, Terraform is ready to make changes to your Okta org. In the Okta Admin Console in your browser, take one last look at the empty Users list (found under the Directory heading in the left sidebar), because you’re about to create some users with Terraform!

In keeping with the Terraform theme, this workshop’s examples will create users for some plants and animals, and assign them to a garden group. However, the exercises will work just as well if you choose a different theme, so feel free to customize your users and groups to keep things interesting.

4.1: Creating your first user

When creating a new Okta object with Terraform, you will always follow the same basic steps:

  • Decide what to create
  • Check the docs for how to create that resource
  • Write Terraform code
  • Plan and apply the Terraform

In this workshop, the first thing you’ll create is a user.

4.2: Write Terraform for a user

Resources in Terraform code describe infrastructure, in this case Okta objects, that Terraform will manage. The first resource you create will describe a user, and applying the Terraform will create that user in the Okta Developer Account whose app integration credentials were passed to the provider configuration.

In the Okta Provider docs on the Terraform Registry, the okta_user resource documentation shows that only the first_name, last_name, login, and email fields are mandatory.

This resource description goes after the provider definition in main.tf:

resource "okta_user" "bird" {
  first_name = "Selasphorus"
  last_name  = "rufous"
  login      = "rufous.hummingbird@example.com"
  email      = "rufous.hummingbird@example.com"
}

Add that resource after the Terraform provider in main.tf. Notice how the type of the resource is okta_user, and the name of this one particular resource is "bird"? The name "bird" is only used within your Terraform code to refer to this particular user. You’ll use it later when adding the user to a group.

4.3: Plan and apply the Terraform

Save main.tf, then run terraform plan in your terraform-workshop directory to see what will change once you run terraform apply:

Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
  + create

Terraform will perform the following actions:

  # okta_user.bird will be created
  + resource "okta_user" "bird" {
      + custom_profile_attributes = (known after apply)
      + email                     = "rufous.hummingbird@example.com"
      + expire_password_on_create = false
      + first_name                = "Selasphorus"
      + id                        = (known after apply)
      + last_name                 = "rufous"
      + login                     = "rufous.hummingbird@example.com"
      + raw_status                = (known after apply)
      + skip_roles                = false
      + status                    = "ACTIVE"
    }

Plan: 1 to add, 0 to change, 0 to destroy.

Should you apply this plan? In this case, you wanted your code to create one user with the specified attributes. The plan indicates that one user with those attributes will be created. If the plan had suggested changing or destroying infrastructure, or creating several objects when you only intended to create one, you would need to go re-assess your code changes instead of applying them.

But there are no surprises in the terraform plan output, so you can apply the changes with terraform apply:

Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
  + create

Terraform will perform the following actions:

  # okta_user.bird will be created
  + resource "okta_user" "bird" {
      + custom_profile_attributes = (known after apply)
      + email                     = "rufous.hummingbird@example.com"
      + expire_password_on_create = false
      + first_name                = "Selasphorus"
      + id                        = (known after apply)
      + last_name                 = "rufous"
      + login                     = "rufous.hummingbird@example.com"
      + raw_status                = (known after apply)
      + skip_roles                = false
      + status                    = "ACTIVE"
    }

Plan: 1 to add, 0 to change, 0 to destroy.
okta_user.bird: Creating...
okta_user.bird: Creation complete after 1s [id=00abcdef12345]

Apply complete! Resources: 1 added, 0 changed, 0 destroyed.

Enter ‘yes’ to confirm that the plan is still what you want, and Terraform creates the Okta user.

In the admin console of your Okta Developer Account, go to Directory in the left column and select People. Do you see the newly created user in the list? Note that the resource name “bird” is only used within Terraform, and doesn’t show up to the user or in the admin interface.

4.4: Create another user

Now you can make a second user with Terraform! Try creating an Okta user resource that Terraform will internally call "butterfly", with the name Papilio zelicaon, and the login and email swallowtail.butterfly@example.com.

Plan and apply your Terraform, and look at the new user in your admin console.

In main.tf, did your user resource look like this?

resource "okta_user" "butterfly" {
  first_name = "Papilio"
  last_name  = "zelicaon"
  login      = "swallowtail.butterfly@example.com"
  email      = "swallowtail.butterfly@example.com"
}

4.5: Create a group

Just like when creating a user, the Okta Terraform Provider docs offer guidance on which fields are available, and which are required, when creating a group.

Configuring the group resource in main.tf will create a group the next time you run terraform apply:

resource "okta_group" "garden" {
  name        = "The Garden"
  description = "Terraform created this group"
}

Save main.tf, and run terraform plan. While the plan runs, think about what changes you expect to see, based on what you’ve changed in your Terraform code since the last time you applied it. Should any resources be added? Should any resources be changed? Should any resources be destroyed?

Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
  + create

Terraform will perform the following actions:

  # okta_group.garden will be created
  + resource "okta_group" "garden" {
      + description = "Terraform created this group"
      + id          = (known after apply)
      + name        = "The Garden"
      + skip_users  = false
    }

Plan: 1 to add, 0 to change, 0 to destroy.

You added 1 new group resource, and Terraform plans to add 1 new group. The plan looks correct. Run terraform apply, enter ‘yes’ if you still like the plan, and then you can see the new group in the admin console.

When you inspect the new group, you may notice that it has no users assigned to it. Let’s fix that!

4.6: Add users to the group

Group memberships are managed with the okta_group_memberships resource. Where would you look for information on what fields an okta_group_memberships resource requires? The provider docs are always a good place to start.

From the provider’s example, you can describe the group membership resource in main.tf:

resource "okta_group_memberships" "gardenmembers" {
  group_id = okta_group.garden.id
  users = [
    okta_user.bird.id,
    okta_user.butterfly.id,
  ]
}

Plan and apply the Terraform, then look at the group in your admin console. Are the garden creatures assigned to the garden group now?

5: Explore Terraform errors

Now that you’re getting more comfortable with basic Terraform operations, let’s pause and look at some ways that things might have gone wrong in the previous steps.

5.1: Multiple keys in the application

To cause this error, navigate to your Terraform application in the Okta Admin Console, and add a second keypair in the Public Keys section of the General configuration tab.

With two keypairs active, try to run terraform plan. Every resource will return an invalid JWT key identifier error: “the API returned an error: The client_assertion JWT kid is invalid.”

Try setting your first keypair, the one you created the RSA key from, to “inactive” in the admin console. This changes the error to “the API returned an error: The client_assertion signature is invalid.”.

To fix this error, set the keypair that Terraform is using to active, and the unused keypair to inactive. Suddenly, the plan succeeds again! Once the unused keypair is set to inactive, you can delete it.

5.2: Missing scopes

To see the impact of a missing scope error, try creating a user when the Terraform provider’s scopes are set up incorrectly. Let’s add a flower to main.tf:

resource "okta_user" "flower" {
  first_name = "Digitalis"
  last_name  = "purpurea"
  login      = "foxglove@example.com"
  email      = "foxglove@example.com"
}

Before planning and applying, go to the Okta API Scopes tab of your app integration in the admin console, and revoke the okta.users.manage scope. When you try to run terraform plan, you’ll get a 403: the API returned an error: The access token provided does not contain the required scopes., Status: 403 Forbidden

Grant that okta.users.manage scope again in the console, and the plan will succeed.

Now in main.tf, in the Okta Terraform provider configuration, try taking away the okta.users.manage scope, so that the scopes line looks like this:

scopes = ["okta.groups.manage", "okta.policies.manage"]

You’ll get the same 403 error as before.

This 403 is your cue to double check that the required scopes are granted in the app integration and also listed in the provider configuration.

When the okta.users.manage scope is granted in both places, you can successfully apply your Terraform to create the foxglove flower.

How would you add the flower to the garden group?

5.3: Resolving clock confusion

For secure communication between Terraform and Okta, the system where you’re running Terraform has to agree with Okta about what time it is.

Try setting the clock of your system to a few minutes in the future, and running a Terraform command.

You’ll see the error the API returned an error: The client_assertion token has an expiration too far into the future.

Fix this error by configuring the system where you’re running Terraform to automatically update its clock from a trusted time server. Forcing your system clock to synchronize to its timeserver can also resolve this issue.

6: Remove resources

Terraform can remove infrastructure as well as create it.

When you delete the Terraform configuration for a managed resource, your next terraform apply will delete that resource.

Hummingbirds can migrate over 2000 miles per year, so let’s remove the hummingbird’s account when it migrates away from the garden.

Simply removing the bird resource description from main.tf and trying to run terraform plan will yield an error:

Error: Reference to undeclared resource

  on main.tf line 123, in resource "okta_group_memberships" "gardenmembers":
 123:     okta_user.bird.id,

A managed resource "okta_user" "bird" has not been declared in the root
module.

What went wrong? The error came from the line where the bird was assigned to the garden group.

To successfully remove a resource, all references to it must also be removed. In this example, the bird resource is only used in its assignment to the garden group. Removing that assignment along with the resource will allow the plan to succeed. Since the resource has the unique name bird, another way to find everywhere it’s referenced would be to search the Terraform project for that string: grep bird *.tf.

In the Okta console, you’ll see that the bird’s user account is not removed until you run terraform apply.

7: Build intermediate Terraform skills

As you use Terraform to describe the complexity of the real world, you’ll need more of its features. In the next steps of this workshop, you’ll practice intermediate Terraform skills as they apply to the Okta provider.

7.1: Use Terraform variables

The simple examples so far have used a lot of hardcoded strings. One in particular is the example.com domain that all of the users have their logins and IDs at. What if you expected that to change, and wanted to move to a new email domain?

Terraform’s variables can help.

7.2: Variable email domain

To declare a variable, in this case a string representing the email domain, tell Terraform the variable type and optionally a default value. This variable definition can go in main.tf while the project is still small:

variable "domain" {
  type    = string
  default = "example.com"
}

Now you can replace all the instances of example.com with var.domain. Here’s how that would look for the butterfly, including the syntax for using the variable within a string:

resource "okta_user" "butterfly" {
  first_name = "Papilio"
  last_name  = "zelicaon"
  login      = "swallowtail.butterfly@${var.domain}"
  email      = "swallowtail.butterfly@${var.domain}"
}

Run terraform plan with no arguments, and the domain variable will use its default of example.com.

If you wanted to change everyone’s domain at once in a big switch from example.com to test.org email hosting, you could pass a different value to the domain variable when running Terraform:

$ terraform plan -var "domain=test.org"
$ terraform apply -var "domain=test.org"

Can you create and use a list variable to replace the hardcoded list in your okta_group_memberships resource?

7.3: Key file management techniques

You’ve already used a built-in Terraform variable to describe the current file path, when telling the Okta provider where to look for the the private key:

  private_key = "${path.module}/rsa.pem"

How else could you have provided that key to your code?

7.4: Import file to Terraform

You could specify the private key with Terraform’s builtin file function:

  private_key = file("rsa.pem")

7.5: Read environment variables from Terraform

If your secrets management solution puts the key file into an environment variable, you can relay it to your Terraform. Here’s how that would look if your Terraform can access the key through an environment variable SECRET_PKEY:

$ terraform plan -var "pk_from_env=${SECRET_PKEY}"

This argument to Terraform reads the SECRET_PKEY environment variable into the pk_from_env Terraform variable. The pk_from_env variable can then be used to tell the Okta provider its private key:

private_key = "${var.pk_from_env}"

This pattern of reading secrets and variables from the shell where Terraform runs is especially common in continuous integration, continuous deployment, and other automation.

7.6: Use multi-line strings

So far you’ve used lists, strings, and string interpolation. What would you do if you needed to set a variable to a value with several lines?

Terraform’s strings support heredoc syntax. Here’s how the garden group looks with a more flowery description:

resource "okta_group" "garden" {
  name        = "The Garden"
  description = <<POEM
When daisies pied and violets blue
And lady-smocks all silver-white
And cuckoo-buds of yellow hue
Do paint the meadows with delight.  
POEM  
}

Notice how << and a custom delimiter, in this case POEM, start the multi-line string. The same delimiter, in this case, POEM on its own line ends the string.

7.7: Refactor Terraform files

As you configure more resources with Terraform, it gets harder to keep track of everything in one big file. Fortunately, Terraform recognizes all .tf files in a directory as belonging to the same project.

To see this in action, try moving both of the remaining user resources, the butterfly and the flower, from main.tf into a new file called users.tf. After this change, terraform plan will report “No changes. Your infrastructure matches the configuration.”

As your Terraform project grows more complex, you’ll use more features of the language. You might use modules to reuse similar configurations. You’ll probably use more variables to take inputs to Terraform from the environment where it’s running.

7.8: Format Terraform code

If you’re using an editor or IDE that doesn’t automatically fix your indentation and other formatting details, run the command terraform fmt from the command line in your Terraform project directory to clean up the formatting of all .tf files.

Try adding some ugly whitespace to one of your Terraform files, and then use terraform fmt to clean it up!

8: Import resources

What if a user existed in Okta before you started working with Terraform, and you wanted to create a Terraform resource to represent them? Terraform import, an experimental feature, does this.

8.1: Manually create users

To try it out, you’ll need something to import. Create a user by hand through the Okta admin console.

Navigate to Directory, People, Add Person. The garden could use a tree in it, so this example will make the tree Acer macrophyllum, bigleaf.maple@example.com.

View that user in the Okta admin console, and its ID appears at the end of the URL: https://dev-1234567890-admin.okta.com/admin/user/profile/view/abc123. Use that ID when importing the resource to the Terraform state.

8.2: Import data with Terraform

Create a new Terraform file, imports.tf, and add an import block to that file:

import {
  to = okta_user.tree
  id = "abc123"
}

Run terraform plan -generate-config-out=generated.tf to generate the configuration for everything defined in import blocks.

After planning, generated.tf contains the value of every possible field on the resource!

# __generated__ by Terraform
# Please review these resources and move them into your main configuration files. 
# __generated__ by Terraform from "abc123"
resource "okta_user" "tree" { 
	city= null
	cost_center = null
	country_code= null
	custom_profile_attributes = "{}"
	custom_profile_attributes_to_ignore = null
	department= null
	display_name= null
	division= null
	email = "bigleaf.maple@example.com" 
	employee_number = null
	expire_password_on_create = null
	first_name= "Acer"
	honorific_prefix= null
	honorific_suffix= null
	last_name = "macrophyllum"
	locale= null
	login = "bigleaf.maple@example.com" 
	manager = null
	manager_id= null
	middle_name = null
	mobile_phone= null
	nick_name = null
	old_password= null # sensitive
	organization= null
	password= null # sensitive
	password_inline_hook= null
	postal_address= null
	preferred_language= null
	primary_phone = null
	profile_url = null
	recovery_answer = null # sensitive
	recovery_question = null
	second_email= null
	state = null
	status= "STAGED"
	street_address= null
	timezone= null
	title = null
	user_type = null
	zip_code= null
}

After the plan creates generated.tf, run terraform apply to complete the import.

8.3: Clean up after import

Copy the fields you’d like to manage from generated.tf into users.tf, and modify them to use any relevant variables. Here’s how the tree might look in main.tf:

resource "okta_user" "tree" {
  email      = "bigleaf.maple@${var.domain}"
  first_name = "Acer"
  last_name  = "macrophyllum"
  login      = "bigleaf.maple@${var.domain}"
}

Once the user is successfully imported, remove imports.tf and generated.tf.

After this refactor, a terraform plan will only show changes that update the manually created user’s settings to the terraform resource defaults.

9: Find and fix configuration drift

Config drift can happen when Okta resources are manually changed through the admin console and no longer match their Terraform configurations. You can introduce configuration drift by manually changing a resource.

9.1: Create configuration drift

Foxgloves are poisonous, so you might not want them in the garden. In the Okta admin console, remove the foxglove from the garden group and suspend their account.

After taking those actions by hand in the web interface, run terraform plan. Terraform detects that the foxglove changed, and wants to put things back to exactly how they’re described in main.tf!

Terraform will perform the following actions:

  # okta_group_memberships.gardenmembers will be updated in-place
  ~ resource "okta_group_memberships" "gardenmembers" {
        id              = "456lmn"
      ~ users           = [
          + "789xyz",
            # (1 unchanged element hidden)
        ]
        # (2 unchanged attributes hidden)
    }

  # okta_user.flower will be updated in-place
  ~ resource "okta_user" "flower" {
        id                        = "z89xyz"
      ~ status                    = "SUSPENDED" -> "ACTIVE"
        # (8 unchanged attributes hidden)
    }

Plan: 0 to add, 2 to change, 0 to destroy.

9.2: Compare configuration drift solutions

The best way for your organization to handle this drift will depend on your business needs. When Terraform and Okta disagree about a resource, it’s up to you to decide which of them is more correct, and change the less correct one so that they match.

9.2.1: Make Terraform match Okta

If manual changes need to be reflected by Terraform changes, use terraform import, as shown above, to change your Terraform.

9.2.2: Fix the config drift

For the purposes of this workshop, assume that Terraform’s configuration represents the desired state of the infrastructure. How should you make Okta match Terraform?

After you run terraform apply, the admin console will show that the foxglove user is back to being active and in the garden group, just how Terraform describes it.

10: Control API usage

Every Okta organization has an API rate limit. You can find the current rate limits in the Okta docs.

If you’re using Terraform in an organization with other API integrations, you might want to fine-tune how much of the organization’s total rate limit you want Terraform to consume. To do this, go to your Terraform application in the admin console, and move the slider on the Application Rate Limits tab there.

11: Clean up

If you would like to remove all resources managed by your Terraform configuration from your Okta Developer account, run the command terraform destroy. This will remove everything that you created during this workshop, so your account is a clean slate next time you want to test your code in it.

Never run terraform destroy on infrastructure that you want to keep!

12: Next steps

In this workshop, you’ve taken your first steps managing Okta with Terraform. You’ve learned to create, change, import, and destroy resources, as well as refactoring and formatting your Terraform code.

Check out the other workshops in the enterprise-ready series if you haven’t already!

Posts in the enterprise-ready workshop series
1. How to Get Going with the Enterprise-Ready SaaS Apps Workshops
2. Enterprise-Ready Workshop: Authenticate with OpenID Connect
3. Enterprise-Ready Workshop: Manage Users with SCIM
4. Enterprise-Ready Workshop: Terraform
5. Enterprise Maturity Workshop: Automate with no-code Okta Workflows
6. How to Instantly Sign a User Out across All Your App

You’ve also begun to consider more advanced topics that you’ll encounter in your Terraform journey, like config drift and API usage.

  • Learn more about Terraform in the Hashicorp docs. Could Terraform help you manage cloud infrastructure beyond Okta?
  • Explore other resource types in the Okta Terraform Provider docs, and try configuring them in your developer account.
  • If your organization already uses Terraform to manageinfrastructure, learn about how terraform.tfstate files and secrets are managed.
  • Leave a comment to let us know what you thought about using Okta’s Terraform module! What features would you like to learn about next?

Follow us on Twitter and subscribe to our YouTube channel to find out about future workshops and tutorials! Leave a comment to let us know what you thought about using Okta’s Terraform module! What features would you like to learn about next?

avatar-edunham.jpg E. Dunham

Okta Developer Blog Comment Policy

We welcome relevant and respectful comments. Off-topic comments may be removed.