• Predictive test jobs before a merge request is approved
  • Overview of the GitLab project test dependency
  • RSpec predictive jobs
    • Determining predictive RSpec test files in a merge request
    • Exceptional cases
  • Jest predictive jobs
    • Determining predictive Jest test files in a merge request
    • Exceptional cases
  • Fork pipelines
• Fail-fast job in merge request pipelines
• Re-run previously failed tests in merge request pipelines
  • How it works?
• Faster feedback for some merge requests
  • Broken Master Fixes
  • Revert MRs
  • The pipeline:expedite label
• Test jobs
  • End-to-end jobs
  • Review app jobs
  • As-if-FOSS jobs
  • As-if-JH cross project downstream pipeline
    • When to consider applying pipeline:run-as-if-jh label
    • Corresponding JH branch
    • How as-if-JH pipeline was configured
      • Tokens set in the project variables
      • How we generate the as-if-JH branch
      • How we trigger and run the as-if-JH pipeline
      • How the GitLab JH mirror project is set up
      • How the GitLab JH validation project is set up
  • rspec:undercoverage job
    • Troubleshooting rspec:undercoverage failures
• Test suite parallelization
• Flaky tests
  • Automatic skipping of flaky tests
  • Automatic retry of failing tests in a separate process
• Compatibility testing
  • Ruby versions testing
  • PostgreSQL versions testing
    • Current versions testing
    • Long-term plan
  • Redis versions testing
    • Current versions testing
  • Single database testing
• Monitoring
• Logging
• Pipelines types for merge requests
  • Documentation pipeline
  • Backend pipeline
  • Frontend pipeline
  • End-to-end pipeline
• CI configuration internals
• Performance

Pipelines for the GitLab project

Pipelines for gitlab-org/gitlab (as well as the dev instance’s) is configured in the usual .gitlab-ci.yml which itself includes files under .gitlab/ci/ for easier maintenance.

We’re striving to dogfood GitLab CI/CD features and best-practices as much as possible.

Predictive test jobs before a merge request is approved

To reduce the pipeline cost and shorten the job duration, before a merge request is approved, the pipeline will run a predictive set of RSpec & Jest tests that are likely to fail for the merge request changes.

After a merge request has been approved, the pipeline would contain the full RSpec & Jest tests. This will ensure that all tests have been run before a merge request is merged.

Overview of the GitLab project test dependency

To understand how the predictive test jobs are executed, we need to understand the dependency between GitLab code (frontend and backend) and the respective tests (Jest and RSpec). This dependency can be visualized in the following diagram:

In summary:

• RSpec tests are dependent on the backend code.
• Jest tests are dependent on both frontend and backend code, the latter through the frontend fixtures.

RSpec predictive jobs

Determining predictive RSpec test files in a merge request

To identify the RSpec tests that are likely to fail in a merge request, we use the test_file_finder gem, with two strategies:

• dynamic mapping from test coverage tracing (generated via the Crystalball gem) (see where it’s used)
• static mapping maintained in the tests.yml file for special cases that cannot be mapped via coverage tracing (see where it’s used)

The test mappings contain a map of each source files to a list of test files which is dependent of the source file.

In the detect-tests job, we use this mapping to identify the predictive tests needed for the current merge request.

Later on in the rspec fail-fast job, we run the predictive tests for the current merge request.

Exceptional cases

In addition, there are a few circumstances where we would always run the full RSpec tests:

• when the pipeline:run-all-rspec label is set on the merge request. This label will trigger all RSpec tests including those run in the as-if-foss jobs.
• when the pipeline:mr-approved label is set on the merge request, and if the code changes satisfy the backend-patterns rule. Note that this label is assigned by triage automation when the merge request is approved by any reviewer. It is not recommended to apply this label manually.
• when the merge request is created by an automation (for example, Gitaly update or MR targeting a stable branch)
• when the merge request is created in a security mirror
• when any CI configuration file is changed (for example, .gitlab-ci.yml or .gitlab/ci/**/*)

Jest predictive jobs

Determining predictive Jest test files in a merge request

To identify the jest tests that are likely to fail in a merge request, we pass a list of all the changed files into jest using the --findRelatedTests option. In this mode, jest would resolve all the dependencies of related to the changed files, which include test files that have these files in the dependency chain.

Exceptional cases

In addition, there are a few circumstances where we would always run the full Jest tests:

• when the pipeline:run-all-jest label is set on the merge request
• when the merge request is created by an automation (for example, Gitaly update or MR targeting a stable branch)
• when the merge request is created in a security mirror
• when relevant CI configuration file is changed (.gitlab/ci/rules.gitlab-ci.yml, .gitlab/ci/frontend.gitlab-ci.yml)
• when any frontend dependency file is changed (for example, package.json, yarn.lock, config/webpack.config.js, config/helpers/**/*.js)
• when any vendored JavaScript file is changed (for example, vendor/assets/javascripts/**/*)

The rules definitions for full Jest tests are defined at .frontend:rules:jest in rules.gitlab-ci.yml.

Fork pipelines

We run only the predictive RSpec & Jest jobs for fork pipelines, unless the pipeline:run-all-rspec label is set on the MR. The goal is to reduce the CI/CD minutes consumed by fork pipelines.

See the experiment issue.

Fail-fast job in merge request pipelines

To provide faster feedback when a merge request breaks existing tests, we are experimenting with a fail-fast mechanism.

An rspec fail-fast job is added in parallel to all other rspec jobs in a merge request pipeline. This job runs the tests that are directly related to the changes in the merge request.

If any of these tests fail, the rspec fail-fast job fails, triggering a fail-pipeline-early job to run. The fail-pipeline-early job:

• Cancels the currently running pipeline and all in-progress jobs.
• Sets pipeline to have status failed.

For example:

The rspec fail-fast is a no-op if there are more than 10 test files related to the merge request. This prevents rspec fail-fast duration from exceeding the average rspec job duration and defeating its purpose.

This number can be overridden by setting a CI/CD variable named RSPEC_FAIL_FAST_TEST_FILE_COUNT_THRESHOLD.

Re-run previously failed tests in merge request pipelines

In order to reduce the feedback time after resolving failed tests for a merge request, the rspec rspec-pg12-rerun-previous-failed-tests and rspec rspec-ee-pg12-rerun-previous-failed-tests jobs run the failed tests from the previous MR pipeline.

This was introduced on August 25th 2021, with https://gitlab.com/gitlab-org/gitlab/-/merge_requests/69053.

How it works?

1. The detect-previous-failed-tests job (prepare stage) detects the test files associated with failed RSpec jobs from the previous MR pipeline.
2. The rspec rspec-pg12-rerun-previous-failed-tests and rspec rspec-ee-pg12-rerun-previous-failed-tests jobs will run the test files gathered by the detect-previous-failed-tests job.

Faster feedback for some merge requests

Broken Master Fixes

When you need to fix a broken master, you can add the pipeline:expedite label to expedite the pipelines that run on the merge request.

Note that the merge request also needs to have the master:broken or master:foss-broken label set.

Revert MRs

To make your Revert MRs faster, use the revert MR template before you create your merge request. It will apply the pipeline:expedite label and others that will expedite the pipelines that run on the merge request.

The pipeline:expedite label

When this label is assigned, the following steps of the CI/CD pipeline are skipped:

• The e2e:package-and-test job.
• The rspec:undercoverage job.
• The entire Review Apps process.

Apply the label to the merge request, and run a new pipeline for the MR.

Test jobs

We have dedicated jobs for each testing level and each job runs depending on the changes made in your merge request. If you want to force all the RSpec jobs to run regardless of your changes, you can add the pipeline:run-all-rspec label to the merge request.

End-to-end jobs

The e2e:package-and-test child pipeline runs end-to-end jobs automatically depending on changes, and is manual in other cases. See .qa:rules:package-and-test in rules.gitlab-ci.yml for the specific list of rules.

If you want to force e2e:package-and-test to run regardless of your changes, you can add the pipeline:run-all-e2e label to the merge request.

Consult the End-to-end Testing dedicated page for more information.

Review app jobs

The start-review-app-pipeline child pipeline deploys a Review App and runs end-to-end tests against it automatically depending on changes, and is manual in other cases. See .review:rules:start-review-app-pipeline in rules.gitlab-ci.yml for the specific list of rules.

If you want to force a Review App to be deployed regardless of your changes, you can add the pipeline:run-review-app label to the merge request.

Consult the Review Apps dedicated page for more information.

As-if-FOSS jobs

The * as-if-foss jobs run the GitLab test suite “as if FOSS”, meaning as if the jobs would run in the context of gitlab-org/gitlab-foss. These jobs are only created in the following cases:

• when the pipeline:run-as-if-foss label is set on the merge request
• when the merge request is created in the gitlab-org/security/gitlab project
• when any CI configuration file is changed (for example, .gitlab-ci.yml or .gitlab/ci/**/*)

The * as-if-foss jobs are run in addition to the regular EE-context jobs. They have the FOSS_ONLY='1' variable set and get the ee/ folder removed before the tests start running.

The intent is to ensure that a change doesn’t introduce a failure after gitlab-org/gitlab is synced to gitlab-org/gitlab-foss.

As-if-JH cross project downstream pipeline

The start-as-if-jh job triggers a cross project downstream pipeline which runs the GitLab test suite “as if JiHu”, meaning as if the pipeline would run in the context of GitLab JH. These jobs are only created in the following cases:

• when the pipeline:run-as-if-jh label is set on the merge request

This pipeline runs under the context of a generated branch in the GitLab JH validation project, which is a mirror of the GitLab JH mirror.

The generated branch name is prefixed with as-if-jh/ along with the branch name in the merge request. This generated branch is based on the merge request branch, additionally adding changes downloaded from the corresponding JH branch on top to turn the whole pipeline as if JiHu.

The intent is to ensure that a change doesn’t introduce a failure after GitLab is synchronized to GitLab JH.

When to consider applying pipeline:run-as-if-jh label

If a Ruby file is renamed and there’s a corresponding prepend_mod line, it’s likely that GitLab JH is relying on it and requires a corresponding change to rename the module or class it’s prepending.

Corresponding JH branch

You can create a corresponding JH branch on GitLab JH by appending -jh to the branch name. If a corresponding JH branch is found, as-if-jh pipeline grabs files from the respective branch, rather than from the default branch main-jh.

How as-if-JH pipeline was configured

The whole process looks like this:

Tokens set in the project variables

• ADD_JH_FILES_TOKEN: This is a GitLab JH mirror project token with read_api permission, to be able to download JiHu files.
• AS_IF_JH_TOKEN: This is a GitLab JH validation project token with write_repository permission, to push generated as-if-jh/* branch.

How we generate the as-if-JH branch

First add-jh-files job will download the required JiHu files from the corresponding JH branch, saving in artifacts. Next prepare-as-if-jh-branch job will create a new branch from the merge request branch, commit the changes, and finally push the branch to the validation project.

How we trigger and run the as-if-JH pipeline

After having the as-if-jh/* branch, start-as-if-jh job will trigger a pipeline in the validation project to run the cross-project downstream pipeline.

How the GitLab JH mirror project is set up

The GitLab JH mirror project is private and CI is disabled.

It’s a pull mirror pulling from GitLab JH, mirroring all branches, overriding divergent refs, triggering no pipelines when mirror is updated.

The pulling user is @gitlab-jh-bot, who is a maintainer in the project. The credentials can be found in the 1password engineering vault.

No password is used from mirroring because GitLab JH is a public project.

How the GitLab JH validation project is set up

This GitLab JH validation project is public and CI is enabled, without any project variables.

It’s a pull mirror pulling from GitLab JH mirror, mirroring only protected branches, master and main-jh, overriding divergent refs, triggering no pipelines when mirror is updated.

The pulling user is @gitlab-jh-validation-bot, who is a maintainer in the project, and also a reporter in the GitLab JH mirror. The credentials can be found in the 1password engineering vault.

A personal access token from @gitlab-jh-validation-bot with write_repository permission is used as the password to pull changes from the GitLab JH mirror. Username is set with gitlab-jh-validation-bot.

There is also a pipeline schedule to run maintenance pipelines with variable SCHEDULE_TYPE set to maintenance running every day, updating cache.

The default CI/CD configuration file is also set at jh/.gitlab-ci.yml so it runs exactly like GitLab JH.

rspec:undercoverage job

The rspec:undercoverage job runs undercover to detect, and fail if any changes introduced in the merge request has zero coverage.

The rspec:undercoverage job obtains coverage data from the rspec:coverage job.

In the event of an emergency, or false positive from this job, add the pipeline:skip-undercoverage label to the merge request to allow this job to fail.

Troubleshooting rspec:undercoverage failures

The rspec:undercoverage job has known bugs that can cause false positive failures. You can test coverage locally to determine if it’s safe to apply pipeline:skip-undercoverage. For example, using <spec> as the name of the test causing the failure:

1. Run SIMPLECOV=1 bundle exec rspec <spec>.
2. Run scripts/undercoverage.

If these commands return undercover: ✅ No coverage is missing in latest changes then you can apply pipeline:skip-undercoverage to bypass pipeline failures.

Test suite parallelization

Our current RSpec tests parallelization setup is as follows:

1. The retrieve-tests-metadata job in the prepare stage ensures we have a knapsack/report-master.json file:
   • The knapsack/report-master.json file is fetched from the latest main pipeline which runs update-tests-metadata (for now it’s the 2-hourly maintenance scheduled master pipeline), if it’s not here we initialize the file with {}.
2. Each [rspec|rspec-ee] [migration|unit|integration|system|geo] n m job are run with knapsack rspec and should have an evenly distributed share of tests:
   • It works because the jobs have access to the knapsack/report-master.json since the “artifacts from all previous stages are passed by default”.
   • the jobs set their own report path to "knapsack/${TEST_TOOL}_${TEST_LEVEL}_${DATABASE}_${CI_NODE_INDEX}_${CI_NODE_TOTAL}_report.json".
   • if knapsack is doing its job, test files that are run should be listed under Report specs, not under Leftover specs.
3. The update-tests-metadata job (which only runs on scheduled pipelines for the canonical project takes all the knapsack/rspec*.json files and merge them all together into a single knapsack/report-master.json file that is saved as artifact.

After that, the next pipeline uses the up-to-date knapsack/report-master.json file.

Flaky tests

Automatic skipping of flaky tests

Tests that are known to be flaky are skipped unless the $SKIP_FLAKY_TESTS_AUTOMATICALLY variable is set to false or if the pipeline:run-flaky-tests label is set on the MR.

See the experiment issue.

Automatic retry of failing tests in a separate process

Unless $RETRY_FAILED_TESTS_IN_NEW_PROCESS variable is set to false (true by default), RSpec tests that failed are automatically retried once in a separate RSpec process. The goal is to get rid of most side-effects from previous tests that may lead to a subsequent test failure.

We keep track of retried tests in the $RETRIED_TESTS_REPORT_FILE file saved as artifact by the rspec:flaky-tests-report job.

See the experiment issue.

Compatibility testing

By default, we run all tests with the versions that runs on GitLab.com.

Other versions (usually one back-compatible version, and one forward-compatible version) should be running in nightly scheduled pipelines.

Exceptions to this general guideline should be motivated and documented.

Ruby versions testing

We’re running Ruby 3.0 for the merge requests and the default branch. However, we’re still running Ruby 2.7 for GitLab.com and there are older versions that we need to maintain, so we also run our test suite against Ruby 2.7 on a dedicated 2-hourly scheduled pipelines.

For merge requests, you can add the pipeline:run-in-ruby2 label to switch the Ruby version used for running the whole test suite to 2.7. When you do this, the test suite will no longer run in Ruby 3.0 (default), and an additional job verify-ruby-3.0 will also run and always fail to remind us to remove the label and run in Ruby 3.0 before merging the merge request.

This should let us:

• Test changes for Ruby 2.7
• Make sure it will not break anything when it’s merged into the default branch

PostgreSQL versions testing

Our test suite runs against PG12 as GitLab.com runs on PG12 and Omnibus defaults to PG12 for new installs and upgrades.

We do run our test suite against PG11 and PG13 on nightly scheduled pipelines.

We also run our test suite against PG11 upon specific database library changes in MRs and main pipelines (with the rspec db-library-code pg11 job).

Current versions testing

There are 2 pipeline schedules used for testing Ruby 2.7. One is triggering a pipeline in ruby2-sync branch, which updates the ruby2 branch with latest master, and no pipelines will be triggered by this push. The other schedule is triggering a pipeline in ruby2 5 minutes after it, which is considered the maintenance schedule to run test suites and update cache.

The ruby2 branch must not have any changes. The branch is only there to set RUBY_VERSION to 2.7 in the maintenance pipeline schedule.

The gitlab job in the ruby2-sync branch uses a gitlab-org/gitlab project token with write_repository scope and Maintainer role with no expiration. The token is stored in the RUBY2_SYNC_TOKEN variable in gitlab-org/gitlab.

Long-term plan

We follow the PostgreSQL versions shipped with Omnibus GitLab:

Redis versions testing

Our test suite runs against Redis 6 as GitLab.com runs on Redis 6 and Omnibus defaults to Redis 6 for new installs and upgrades.

We do run our test suite against Redis 5 on nightly scheduled pipelines, specifically when running backward-compatible and forward-compatible PostgreSQL jobs.

Current versions testing

Single database testing

By default, all tests run with multiple databases.

We also run tests with a single database in nightly scheduled pipelines, and in merge requests that touch database-related files.

If you want to force tests to run with a single database, you can add the pipeline:run-single-db label to the merge request.

Monitoring

The GitLab test suite is monitored for the main branch, and any branch that includes rspec-profile in their name.

Logging

• Rails logging to log/test.log is disabled by default in CI for performance reasons. To override this setting, provide the RAILS_ENABLE_TEST_LOG environment variable.

Pipelines types for merge requests

In general, pipelines for an MR fall into one of the following types (from shorter to longer), depending on the changes made in the MR:

• Documentation pipeline: For MRs that touch documentation.
• Backend pipeline: For MRs that touch backend code.
• Frontend pipeline: For MRs that touch frontend code.
• End-to-end pipeline: For MRs that touch code in the qa/ folder.

A “pipeline type” is an abstract term that mostly describes the “critical path” (for example, the chain of jobs for which the sum of individual duration equals the pipeline’s duration). We use these “pipeline types” in metrics dashboards to detect what types and jobs need to be optimized first.

An MR that touches multiple areas would be associated with the longest type applicable. For instance, an MR that touches backend and frontend would fall into the “Frontend” pipeline type since this type takes longer to finish than the “Backend” pipeline type.

We use the rules: and needs: keywords extensively to determine the jobs that need to be run in a pipeline. Note that an MR that includes multiple types of changes would have a pipelines that include jobs from multiple types (for example, a combination of docs-only and code-only pipelines).

Following are graphs of the critical paths for each pipeline type. Jobs that aren’t part of the critical path are omitted.

Documentation pipeline

Reference pipeline.

Backend pipeline

Reference pipeline.

Frontend pipeline

Reference pipeline.

End-to-end pipeline

Reference pipeline.

CI configuration internals

See the dedicated CI configuration internals page.

Performance

See the dedicated CI configuration performance page.

Return to Development documentation