Testing with feature flags

To run a specific test with a feature flag enabled you can use the QA::Runtime::Feature class to enable and disable feature flags (via the API).

Note that administrator authorization is required to change feature flags. QA::Runtime::Feature automatically authenticates as an administrator as long as you provide an appropriate access token via GITLAB_QA_ADMIN_ACCESS_TOKEN (recommended), or provide GITLAB_ADMIN_USERNAME and GITLAB_ADMIN_PASSWORD.

Please be sure to include the tag :requires_admin so that the test can be skipped in environments where administrator access is not available.

WARNING: You are strongly advised to enable feature flags only for a group, project, user, or feature group. This makes it possible to test a feature in a shared environment without affecting other users.

For example, the code below would enable a feature flag named :feature_flag_name for the project created by the test:

RSpec.describe "with feature flag enabled", :requires_admin do
  let(:project) { Resource::Project.fabricate_via_api! }

  before do
    Runtime::Feature.enable(:feature_flag_name, project: project)

  it "feature flag test" do
    # Execute the test with the feature flag enabled.
    # It will only affect the project created in this test.

  after do
    Runtime::Feature.disable(:feature_flag_name, project: project)

Note that the enable and disable methods first set the flag and then check that the updated value is returned by the API.

Similarly, you can enable a feature for a group, user, or feature group:

group = Resource::Group.fabricate_via_api!
Runtime::Feature.enable(:feature_flag_name, group: group)

user = Resource::User.fabricate_via_api!
Runtime::Feature.enable(:feature_flag_name, user: user)

feature_group = "a_feature_group"
Runtime::Feature.enable(:feature_flag_name, feature_group: feature_group)

If no scope is provided, the feature flag is set instance-wide:

# This will affect all users!

Working with selectors

A new feature often replaces a vue component or a haml file with a new one. In most cases, the new file or component is accessible only with a feature flag. This approach becomes problematic when tests must pass both with, and without, the feature flag enabled. To ensure tests pass in both scenarios:

  1. Create another selector inside the new component or file.
  2. Give it the same name as the old one.

Selectors are connected to a specific frontend file in the page object, and checked for availability inside our qa:selectors test. If the mentioned selector is missing inside that frontend file, the test fails. To ensure selectors are available when a feature flag is enabled or disabled, add the new selector to the page object, leaving the old selector in place. The test uses the correct selector and still detects missing selectors.

If a new feature changes an existing frontend file that already has a selector, you can add a new selector with the same name. However, only one of the selectors displays on the page. You should:

  1. Disable the other with the feature flag.
  2. Add a comment in the frontend file to delete the old selector from the frontend file and from the page object file when the feature flag is removed.

Example before

# This is the link to the old file
view 'app/views/devise/passwords/edit.html.haml' do
  # The new selector should have the same name
  element :password_field

Example after

view 'app/views/devise/passwords/edit.html.haml' do
  element :password_field

# Now it can verify the selector is available
view 'app/views/devise/passwords/new_edit_behind_ff.html.haml' do
  # The selector has the same name
  element :password_field

Running a scenario with a feature flag enabled

It's also possible to run an entire scenario with a feature flag enabled, without having to edit existing tests or write new ones.

Please see the QA README for details.

Confirming that end-to-end tests pass with a feature flag enabled

End-to-end tests should pass with a feature flag enabled before it is enabled on Staging or on GitLab.com. Tests that need to be updated should be identified as part of quad-planning. The relevant counterpart Software Engineer in Test is responsible for updating the tests or assisting another engineer to do so. However, if a change does not go through quad-planning and a required test update is not made, test failures could block deployment.

Automatic test execution when a feature flag definition changes

If a merge request adds or edits a feature flag definition file, two package-and-qa jobs will be included automatically in the merge request pipeline. One job will enable the defined feature flag and the other will disable it. The jobs execute the same suite of tests to confirm that they pass with if the feature flag is either enabled or disabled.

Test execution during feature development

If an end-to-end test enables a feature flag, the end-to-end test suite can be used to test changes in a merge request by running the package-and-qa job in the merge request pipeline. If the feature flag and relevant changes have already been merged, you can confirm that the tests pass on the default branch. The end-to-end tests run on the default branch every two hours, and the results are posted to a Test Session Report, which is available in the testcase-sessions project.

If the relevant tests do not enable the feature flag themselves, you can check if the tests will need to be updated by opening a draft merge request that enables the flag by default via a feature flag definition file. That will automatically execute the end-to-end test suite. The merge request can be closed once the tests pass. If you need assistance to update the tests, please contact the relevant stable counterpart in the Quality department, or any Software Engineer in Test if there is no stable counterpart for your group.