Google Cloud Workload Identity Federation and IAM policies

Tier: Free, Premium, Ultimate Offering: GitLab.com
History

To use Google Cloud integrations like the Google Artifact Management integration, you must create and configure a workload identity pool and provider. The Google Cloud integration uses Workload Identity Federation to grant GitLab workloads access to Google Cloud resources through OpenID Connect (OIDC) by using JSON Web Token (JWT) tokens.

Workload Identity Federation

Workload Identity Federation lets you use Identity and Access Management (IAM) to grant external identities IAM roles.

Traditionally, applications running outside Google Cloud used service account keys to access Google Cloud resources. However, service account keys are powerful credentials, and can present a security risk if they are not managed correctly.

With identity federation, you can use Identity and Access Management (IAM) to grant external identities IAM roles directly, without requiring service accounts. This approach eliminates the maintenance and security burden associated with service accounts and their keys.

Workload identity pools

A workload identity pool is an entity that lets you manage non-Google identities on Google Cloud.

The GitLab on Google Cloud integration walks you through setting up a workload identity pool to authenticate to Google Cloud. This setup includes mapping your GitLab role attributes to IAM claims in your Google Cloud IAM policy. For a full list of available GitLab attributes for the GitLab on Google Cloud integration, see OIDC custom claims.

Workload identity pool providers

A workload identity pool provider is an entity that describes a relationship between Google Cloud and your Identity provider (IdP). GitLab is the IdP for your workload identity pool for the GitLab on Google Cloud integration.

For more information on identity federation for external workloads, see Workload Identity Federation.

The default GitLab on Google Cloud integration assumes you want to set up your authentication from GitLab to Google Cloud at the GitLab organization level. If you want to control access to Google Cloud on a per project basis, then you must configure your IAM policies for your workload identity pool provider. For more information on controlling who can access Google Cloud from your GitLab organization, see Access control with IAM.

GitLab authentication with Workload Identity Federation

After your workload identity pool and provider are set up to map your GitLab roles and permissions to IAM roles, you can provision runners to deploy workloads from GitLab to Google Cloud by setting the identity keyword to google_cloud for authorization on Google Cloud.

For more information on provisioning runners using the GitLab on Google Cloud integration, see the tutorial Provisioning runners in Google Cloud.

Create and configure a Workload Identity Federation

To set up the Workload Identity Federation you can either:

  • Use the GitLab UI for a guided setup.
  • Use the Google Cloud CLI to set up the Workload Identity Federation manually.

With the GitLab UI

To use the GitLab UI to set up the Workload Identity Federation:

  1. On the left sidebar, select Search or go to and find your project.
  2. Select Settings > Integrations.
  3. Locate the Google Cloud IAM integration and select Configure.
  4. Select Guided setup and follow the instructions.
note
Due to a known issue, the fields in the page for the Google Cloud IAM integration might not populate after you run the script in the guided setup. If the fields are empty, refresh the page. For more information, see issue 448831.

With the Google Cloud CLI

Prerequisites:

  1. Create a workload identity pool with the following command. Replace these values:

    • <your_google_cloud_project_id> with your Google Cloud project ID. To improve security, use a dedicated project for identity management, separate from resources and CI/CD projects.
    • <your_identity_pool_id> with the ID to use for the pool, which must be 4 to 32 lowercase letters, digits, or hyphens. To avoid collisions, use a unique ID. You should include the GitLab project ID or project path as it facilitates IAM policy management. For example, gitlab-my-project-name.
    gcloud iam workload-identity-pools create <your_identity_pool_id> \
             --project="<your_google_cloud_project_id>" \
             --location="global" \
             --display-name="Workload identity pool for GitLab project ID"
    
  2. Add an OIDC provider to the workload identity pool with the following command. Replace these values:

    • <your_identity_provider_id> with the ID to use for the provider, which must be 4 to 32 lowercase letters, digits, or hyphens. To avoid collisions, use a unique ID in the identity pool. For example, gitlab.
    • <your_google_cloud_project_id> with your Google Cloud project ID.
    • <your_identity_pool_id> with the ID of the workload identity pool you created in the previous step.
    • <your_issuer_uri> with your identity provider issuer URI, which can be can be copied from the IAM integration page when choosing manual setup and must exactly match the value. The parameter must include the path of the root group. For example, if the project is under my-root-group/my-sub-group/project-a, the issuer-uri must be set to https://auth.gcp.gitlab.com/oidc/my-root-group.
    gcloud iam workload-identity-pools providers create-oidc "<your_identity_provider_id>" \
          --location="global" \
          --project="<your_google_cloud_project_id>" \
          --workload-identity-pool="<your_identity_pool_id>" \
          --issuer-uri="<your_issuer_uri>" \
          --display-name="GitLab OIDC provider" \
          --attribute-mapping="attribute.guest_access=assertion.guest_access,\
    attribute.reporter_access=assertion.reporter_access,\
    attribute.developer_access=assertion.developer_access,\
    attribute.maintainer_access=assertion.maintainer_access,\
    attribute.owner_access=assertion.owner_access,\
    attribute.namespace_id=assertion.namespace_id,\
    attribute.namespace_path=assertion.namespace_path,\
    attribute.project_id=assertion.project_id,\
    attribute.project_path=assertion.project_path,\
    attribute.user_id=assertion.user_id,\
    attribute.user_login=assertion.user_login,\
    attribute.user_email=assertion.user_email,\
    attribute.user_access_level=assertion.user_access_level,\
    google.subject=assertion.sub"
    
  • The attribute-mapping parameter must include the mapping between OIDC custom claims included in the JWT ID token to the corresponding identity attributes that are used in Identity and Access Management (IAM) policies to grant access. For more information, see the supported OIDC custom claims that you can use to control access to Google Cloud.

To restrict identity token access to a specific GitLab project or group, use an attribute condition. Use the attribute assertion.project_id for a project and the attribute assertion.namespace_id for a group. For more information, see the Google Cloud documentation about how to define an attribute condition. After you define the attribute condition, you can update the workload identity provider.

After you create the workload identity pool and provider, to complete the setup in GitLab:

  1. On the left sidebar, select Search or go to and find your project.
  2. Select Settings > Integrations.
  3. Locate the Google Cloud IAM integration and select Configure.
  4. Select Manual setup
  5. Complete the fields.
    • Project ID for the Google Cloud project in which you created the workload identity. pool and provider. Example: my-sample-project-191923.
    • Project number for the same Google Cloud project. Example: 314053285323.
    • Pool ID of the workload identity pool you created for this integration.
    • Provider ID of the workload identity provider you created for this integration.

OIDC custom claims

The ID token includes the following custom claims:

Claim name When Description
namespace_id On project events ID of the group or user level namespace.
namespace_path On project events Path of the group or user level namespace.
project_id On project events ID of the project.
project_path On project events Path of the project.
root_namespace_id On group events ID of the root group or user level namespace.
root_namespace_path On group events Path of the root group or user level namespace.
user_id On user-trigged events ID of the user.
user_login On user-trigged events Username of the user.
user_email On user-trigged events Email of the user.
ci_config_ref_uri During CI/CD pipeline run The ref path to the top-level CI pipeline definition.
ci_config_sha During CI/CD pipeline run Git commit SHA for the ci_config_ref_uri.
job_id During CI/CD pipeline run ID of the CI job.
pipeline_id During CI/CD pipeline run ID of the CI pipeline.
pipeline_source During CI/CD pipeline run CI pipeline source.
project_visibility During CI/CD pipeline run The visibility of the project where the pipeline is running.
ref During CI/CD pipeline run Git ref for the CI job.
ref_path During CI/CD pipeline run Fully qualified ref for the CI job.
ref_protected During CI/CD pipeline run If the Git ref is protected.
ref_type During CI/CD pipeline run Git ref type.
runner_environment During CI/CD pipeline run The type of runner used by the CI job.
runner_id During CI/CD pipeline run ID of the runner executing the CI job.
sha During CI/CD pipeline run The commit SHA for the CI job.
environment During CI/CD pipeline run Environment the CI job deploys to.
environment_protected During CI/CD pipeline run If deployed environment is protected.
environment_action During CI/CD pipeline run Environment action specified in the CI job.
deployment_tier During CI/CD pipeline run Deployment tier of the environment the CI job specifies.
user_access_level On user-trigged events Role of the user with values of guest, reporter, developer, maintainer, owner.
guest_access On user-trigged events Indicates whether the user has at least guest role, with values of “true” or “false” as a string.
reporter_access On user-trigged events Indicates whether the user has at least reporter role, with values of “true” or “false” as a string.
developer_access On user-trigged events Indicates whether the user has at least developer role, with values of “true” or “false” as a string.
maintainer_access On user-trigged events Indicates whether the user has at least maintainer role, with values of “true” or “false” as a string.
owner_access On user-trigged events Indicates whether the user has at least owner role, with values of “true” or “false” as a string.

These claims are a superset of the ID token claims. All values are of type string. See the ID token claims documentation for more details and example values.