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.

Do not use CI/CD components in gitlab-org/gitlab pipelines unless they are mirrored on the dev.gitlab.com instance. CI/CD components do not work across different instances, and cause failing pipelines on the dev.gitlab.com mirror if they do not exist on that instance.

Pipeline tiers

A merge request will typically run several CI/CD pipelines. Depending on where the merge request is at in the approval process, we will trigger different kinds of pipelines. We call those kinds of pipelines pipeline tiers.

We currently have three tiers:

1. pipeline::tier-1: The merge request has no approvals
2. pipeline::tier-2: The merge request has at least one approval, but still requires more approvals
3. pipeline::tier-3: The merge request has all the approvals it needs

Typically, the lower the pipeline tier, the fastest the pipeline should be. The higher the pipeline tier, the more confidence the pipeline should give us by running more tests

See the Introduce "tiers" in MR pipelines epic for more information on the implementation.

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:

flowchart LR
    subgraph frontend
    fe["Frontend code"]--tested with-->jest
    end
    subgraph backend
    be["Backend code"]--tested with-->rspec
    end

    be--generates-->fixtures["frontend fixtures"]
    fixtures--used in-->jest

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.

The detect-tests CI job

Most CI/CD pipelines for gitlab-org/gitlab will run a detect-tests CI job in the prepare stage to detect which backend/frontend tests should be run based on the files that changed in the given MR.

The detect-tests job will create many files that will contain the backend/frontend tests that should be run. Those files will be read in subsequent jobs in the pipeline, and only those tests will be executed.

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 dynamic mappings and static mappings.

Dynamic mappings

First, we use the test_file_finder gem, with dynamic mapping strategies coming from the Crystalball gem) (see where it's used, and the mapping strategies we use in Crystalball).

In addition to test_file_finder, we have added several advanced mappings to detect even more tests to run:

• FindChanges (!74003)
  • Automatically detect Jest tests to run upon backend changes (via frontend fixtures)
• PartialToViewsMappings (#395016)
  • Run view specs when Rails partials included in those views are changed in an MR
• JsToSystemSpecsMappings (#386754)
  • Run certain system specs if a JavaScript file was changed in an MR
• GraphqlBaseTypeMappings (#386756)
  • If a GraphQL type class changed, we should try to identify the other GraphQL types that potentially include this type, and run their specs.
• ViewToSystemSpecsMappings (#395017)
  • When a view gets changed, we try to find feature specs that would test that area of the code.
• ViewToJsMappings (#386719)
  • If a JS file is changed, we should try to identify the system specs that are covering this JS component.
• FindFilesUsingFeatureFlags (#407366)
  • If a feature flag was changed, we check which Ruby file is including that feature flag, and we add it to the list of changed files in the detect-tests CI job. The remainder of the job will then detect which frontend/backend tests should be run based on those changed files.

Static mappings

We use the test_file_finder gem, with a static mapping maintained in the tests.yml file for special cases that cannot be mapped via dynamic mappings (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.

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/**/*)

Have you encountered a problem with backend predictive tests?

If so, have a look at the Development Analytics RUNBOOK on predictive tests for instructions on how to act upon predictive tests issues. Additionally, if you identified any test selection gaps, let @gl-dx/development-analytics know so that we can take the necessary steps to optimize test selections.

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.

Have you encountered a problem with frontend predictive tests?

If so, have a look at the Development analytics RUNBOOK on predictive tests for instructions on how to act upon predictive tests issues.

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 compute quota 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 implemented 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:

graph LR
    subgraph "prepare stage";
        A["detect-tests"]
    end

    subgraph "test stage";
        B["jest"];
        C["rspec migration"];
        D["rspec unit"];
        E["rspec integration"];
        F["rspec system"];
        G["rspec fail-fast"];
    end

    subgraph "post-test stage";
        Z["fail-pipeline-early"];
    end

    A --"artifact: list of test files"--> G
    G --"on failure"--> Z

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-pg17-rerun-previous-failed-tests and rspec rspec-ee-pg17-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 the failed test is re-run

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-pg17-rerun-previous-failed-tests and rspec rspec-ee-pg17-rerun-previous-failed-tests jobs will run the test files gathered by the detect-previous-failed-tests job.

graph LR
    subgraph "prepare stage";
        A["detect-previous-failed-tests"]
    end

    subgraph "test stage";
        B["rspec rspec-pg17-rerun-previous-failed-tests"];
        C["rspec rspec-ee-pg17-rerun-previous-failed-tests"];
    end

    A --"artifact: list of test files"--> B & C

Merge trains

Current usage

We started using merge trains in June 2024.

At the moment, Merge train pipelines don't run any tests: they only enforce the "Merging a merge request" guidelines that already existed before the enablement of merge trains, but that we couldn't easily enforce.

Merge train pipelines run a single pre-merge-checks job which ensures the latest pipeline before merge is:

1. A Merged Results pipeline
2. A tier-3 pipeline (a full pipeline, not a predictive one)
3. Created at most 16 hours ago (72 hours for stable branches)

We opened a feedback issue to iterate on this solution.

Next iterations

We opened a dedicated issue to discuss the next iteration for merge trains to actually start running tests in merge train pipelines.

Challenges for enabling merge trains running "full" test pipelines

Why do we need to have a "stable" default branch?

If the default branch is unstable (for example, the CI/CD pipelines for the default branch are failing frequently), all of the merge requests pipelines that were added AFTER a faulty merge request pipeline would have to be canceled and added back to the train, which would create a lot of delays if the merge train is long.

How stable does the default branch have to be?

We don't have a specific number, but we need to have better numbers for flaky tests failures and infrastructure failures (see the Master Broken Incidents RCA Dashboard).

Faster feedback for some merge requests

Broken master Fixes

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

The pipeline::expedited label

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

• The e2e:test-on-omnibus-ee job.
• The rspec:undercoverage job.

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.

[!warning] Forcing all jobs on docs only related MRs would not have the prerequisite jobs and would lead to errors

End-to-end jobs

For more information, see End-to-end test pipelines.

Observability end-to-end jobs

The GitLab Observability Backend has dedicated end-to-end tests that run against a GitLab instance. These tests are designed to ensure the integration between GitLab and the Observability Backend is functioning correctly.

The GitLab pipeline has dedicated jobs (see observability-backend.gitlab-ci.yml) that can be executed from GitLab MRs. These jobs will trigger the E2E tests on the GitLab Observability Backend pipeline against a GitLab instance built from the GitLab MR branch. These jobs are useful to make sure that the GitLab changes under review will not break E2E tests on the GitLab Observability Backend pipeline.

There are two Observability end-to-end jobs:

• e2e:observability-backend-main-branch: executes the tests against the main branch of the GitLab Observability Backend.
• e2e:observability-backend: executes the tests against a branch of the GitLab Observability Backend with the same name as the MR branch.

The Observability E2E jobs are triggered automatically only for merge requests that touch relevant files, such as those in the lib/gitlab/observability/ directory or specific configuration files related to observability features.

To run these jobs manually, you can add the pipeline:run-observability-e2e-tests-main-branch or pipeline:run-observability-e2e-tests-current-branch label to your merge request.

In the following example workflow, a developer creates an MR that touches Observability code and uses Observability end-to-end jobs:

1. A developer creates a GitLab MR that touches observability code. The MR automatically executes the e2e:observability-backend-main-branch job.
2. If e2e:observability-backend-main-branch fails, it means that either the MR broke something (and needs fixing), or the MR made changes that requires the e2e tests to be updated.
3. To update the e2e tests, the developer should:
   1. Create a branch in the GitLab Observability Backend repository, with the same name as the GitLab branch containing the breaking changes.
   2. Fix the e2e tests.
   3. Create a merge request with the changes.
4. The developer should add the pipeline:run-observability-e2e-tests-current-branch label on the GitLab MR and wait for the e2e:observability-backend job to succeed.
5. If e2e:observability-backend succeeds, the developer can merge both MRs.

In addition, the developer can manually add pipeline:run-observability-e2e-tests-main-branch to force the MR to run the e2e:observability-backend-main-branch job. This could be useful in case of changes to files that are not being tracked as related to observability.

There might be situations where the developer would need to skip those tests. To skip tests:

• For an MR, apply the label pipeline:skip-observability-e2e-tests label.
• For a whole project, set the CI variable SKIP_GITLAB_OBSERVABILITY_BACKEND_TRIGGER.

As-if-FOSS jobs and cross project downstream pipeline

To ensure the relevant changes are working properly in the FOSS project, under some conditions we also run:

• * as-if-foss jobs in the same pipeline
• Cross project downstream FOSS pipeline

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. On the other hand, cross project downstream FOSS pipeline actually runs inside the FOSS project, which should be even closer to an actual FOSS environment.

We run them in the following cases:

• when the pipeline:run-as-if-foss label is set on the merge request
• when the pipeline:as-if-foss-run-predictive label is set on the merge request (runs only predictive tests, RuboCop, eslint, and static analysis in the FOSS pipeline)
• when the merge request is created in the gitlab-org/security/gitlab project
• when 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.

Cross project downstream FOSS pipeline simulates merging the merge request into the default branch in the FOSS project instead, which removes a list of files. The list can be found in .gitlab/ci/as-if-foss.gitlab-ci.yml and in merge-train/bin/merge-train.

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

Tokens set in the project variables

• AS_IF_FOSS_TOKEN: This is a GitLab FOSS project token with developer role and write_repository permission, to push generated as-if-foss/* branch.
  • Note that the same name for the security project should use another token from the security FOSS project, so that we never push security changes to a public project.

As-if-JH cross project downstream pipeline

What it is

This pipeline is also called JiHu validation pipeline, and it's currently allowed to fail. When that happens, follow What to do when the validation pipeline fails.

How we run it

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 changes are made to feature flags
• 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.

For now, CI will try to fetch the branch on the GitLab JH mirror, so it might take some time for the new JH branch to propagate to the mirror.

[!note] While GitLab JH validation is a mirror of GitLab JH mirror, it does not include any corresponding JH branch beside the default main-jh. This is why when we want to fetch corresponding JH branch we should fetch it from the main mirror, rather than the validation project.

How as-if-JH pipeline was configured

The whole process looks like this:

[!note] We only run sync-as-if-jh-branch when there are dependencies changes.

flowchart TD
  subgraph "JiHuLab.com"
    JH["gitlab-cn/gitlab"]
  end

  subgraph "GitLab.com"
    Mirror["gitlab-org/gitlab-jh-mirrors/gitlab"]

    subgraph MR["gitlab-org/gitlab merge request"]
      Add["add-jh-files job"]
      Prepare["prepare-as-if-jh-branch job"]
      Add --"download artifacts"--> Prepare
    end

    subgraph "gitlab-org-sandbox/gitlab-jh-validation"
      Sync["(*optional) sync-as-if-jh-branch job on branch as-if-jh-code-sync"]
      Start["start-as-if-jh job on as-if-jh/* branch"]
      AsIfJH["as-if-jh pipeline"]
    end

    Mirror --"pull mirror with master and main-jh"--> gitlab-org-sandbox/gitlab-jh-validation
    Mirror --"download JiHu files with ADD_JH_FILES_TOKEN"--> Add
    Prepare --"push as-if-jh branches with AS_IF_JH_TOKEN"--> Sync
    Sync --"push as-if-jh branches with AS_IF_JH_TOKEN"--> Start
    Start --> AsIfJH
  end

  JH --"pull mirror with corresponding JH branches"--> Mirror

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 developer role and 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.

Optionally, if the merge requests have changes to the dependencies, we have an additional step to run sync-as-if-jh-branch job to trigger a downstream pipeline on as-if-jh-code-sync branch in the validation project. This job will perform the same process as JiHu code-sync, making sure the dependencies changes can be brought to the as-if-jh branch prior to run the validation pipeline.

If there are no dependencies changes, we don't run this process.

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

After having the as-if-jh/* branch prepared and optionally synchronized, 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-validation-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, with temporary project variables set.

It's a pull mirror pulling from GitLab JH mirror, mirroring specific branches: (master|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 maintainer 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.

Additionally, a special branch as-if-jh-code-sync is set and protected. Maintainers can push and developers can merge for this branch. We need to set it so developers can merge because we need to let developers to trigger pipelines for this branch. This is a compromise before we resolve Developer-level users no longer able to run pipelines on protected branches.

It's used to run sync-as-if-jh-branch to synchronize the dependencies when the merge requests changed the dependencies. See How we generate the as-if-JH branch for its implementation.

Temporary GitLab JH validation project variables

• BUNDLER_CHECKSUM_VERIFICATION_OPT_IN is set to false
  • We can remove this variable after JiHu has jh/Gemfile.checksum committed. More context can be found at: Setting it to false to skip it

Why do we have both the mirror project and validation project?

We have separate projects for a several reasons.

• Security: Previously, we had the mirror project only. However, to fully mitigate a security issue, we had to make the mirror project private.

• Isolation: We want to run JH code in a completely isolated and standalone project. We should not run it under the gitlab-org group, which is where the mirror project is. The validation project is completely isolated.

• Cost: We don't want to connect to JiHuLab.com from each merge request. It is more cost effective to mirror the code from JiHuLab.com to somewhere at GitLab.com, and have our merge requests fetch code from there. This means that the validation project can fetch code from the mirror, rather than from JiHuLab.com. The mirror project will periodically fetch from JiHuLab.com.

• Branch separation/security/efficiency: We want to mirror all branches, so that we can fetch the corresponding JH branch from JiHuLab.com. However, we don't want to overwrite the as-if-jh-code-sync branch in the validation project, because we use it to control the validation pipeline and it has access to AS_IF_JH_TOKEN. However, we cannot mirror all branches except a single one. See this issue for details.

  Given this issue, the validation project is set to only mirror master and main-jh. Technically, we don't even need those branches, but we do want to keep the repository up-to-date with all the default branches so that when we push changes from the merge request, we only need to push changes from the merge request, which can be more efficient.

• Separation of concerns:

  • Validation project only has the following branches:
    • master and main-jh to keep changes up-to-date.
    • as-if-jh-code-sync for dependency synchronization. We should never mirror this.
    • as-if-jh/* branches from the merge requests. We should never mirror these.
  • All branches from the mirror project are all coming from JiHuLab.com. We never push anything to the mirror project, nor does it run any pipelines. CI/CD is disabled in the mirror project.

We can consider merging the two projects to simplify the setup and process, but we need to make sure that all of these reasons are no longer concerns.

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.

If the rspec:undercoverage job detects missing coverage due to a CE method being overridden in EE, add the pipeline:run-as-if-foss label to the merge request and start a new pipeline.

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

First, review the coverage data from the gitlab.lcov artifact from the rspec:coverage job. The rspec:coverage job might have failed to gather coverage data for various reasons.

It is possible the rspec:undercoverage job will detect undercoverage for a method call but does not display warnings.

loc: app/controllers/projects/attestations_controller.rb:84:102, coverage: 87.5%
  def parsed_attestation_file hits: n/a
    @parsed_attestation_file ||= begin hits: 52
      if attestation && attestation_file hits: 12 branches: 1/1
        Gitlab::Json.parse(attestation_file.read) hits: 10
      else hits: n/a
        {} hits: 2
      end hits: n/a
    rescue JSON::ParserError => e hits: n/a
      Gitlab::AppJsonLogger.error( hits: 2
        message: 'Failed to parse attestation file', hits: n/a
        error_class: e.class.name, hits: n/a
        error_message: e.message, hits: n/a
        attestation_id: attestation&.id, hits: n/a
        project_id: project.id, hits: n/a
        feature_category: 'artifact_security' hits: n/a
      ) hits: n/a
      {} hits: 2
    end hits: n/a
  end hits: n/a

The rspec:undercoverage job has known bugs that can cause false positive failures. Such false positive failures may also happen if you are updating database migration that is too old. 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 RUN_ALL_MIGRATION_TESTS=1 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.

If you have to use pipeline:skip-undercoverage to bypass unrelated pipeline failures, please open a follow up MR to add the required test coverage. This leaves the system in a better state for others.

pajamas_adoption job

{{< history >}}

• Introduced in GitLab 16.8.

{{< /history >}}

The pajamas_adoption job runs the Pajamas Adoption Scanner in merge requests to prevent regressions in the adoption of the Pajamas Design System.

The job fails if the scanner detects regressions caused by a merge request. If the regressions cannot be fixed in the merge request, add the pipeline:skip-pajamas-adoption label to the merge request, then retry the job.

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 and updates the knapsack/report-master.json in 2 ways:
   1. By default, it 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.
   2. (Experimental) When the AVERAGE_KNAPSACK_REPORT environment variable is set to true, instead of merging the reports, the job will calculate the average of the test duration between knapsack/report-master.json and knapsack/rspec*.json to reduce the performance impact from potentially random factors such as spec ordering, runner hardware differences, flaky tests, etc. This experimental approach is aimed to better predict the duration for each spec files to distribute load among parallel jobs more evenly so the jobs can finish around the same time.

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

Code coverage

We collect code coverage data from our test suites to power test selection, coverage analytics, and flaky test analysis.

Coverage types

CI jobs

Coverage data flows through several CI jobs:

1. Collection: Tests run with coverage instrumentation

   • rspec jobs collect backend coverage via SimpleCov
   • jest jobs collect frontend coverage via Istanbul
   • e2e:test-on-gdk collects E2E coverage via Coverband (backend) and Istanbul (frontend)
   • workhorse jobs collect Go coverage

2. Merging: Coverage from parallel jobs and E2E is merged

   • rspec:coverage merges RSpec coverage into coverage/lcov/gitlab.lcov
   • coverage-frontend merges Jest coverage
   • Merge scripts combine E2E coverage with unit/integration coverage

3. Export: Merged coverage is exported to ClickHouse

   • test-coverage:export-rspec-and-e2e exports backend coverage
   • test-coverage:export-jest-and-e2e exports frontend coverage
   • test-coverage:export-workhorse exports Workhorse coverage

For detailed documentation on coverage collection, data flow, and ClickHouse storage, see Code coverage.

Flaky tests retry and quarantine

Automatic retry of failing tests

Backend tests that fail are automatically retried once in a separate RSpec process. This helps detect flaky tests, defined as tests that fail then pass with the same commit SHA.

The "retrying failed tests in a new RSpec process" can be disabled by setting the $RETRY_FAILED_TESTS_IN_NEW_PROCESS variable to false.

Quarantined tests

GitLab CI pipelines use fast quarantine to skip tests that are blocking pipelines while being investigated.

The fast quarantine process can be disabled by setting the $FAST_QUARANTINE variable to false.

For quarantine procedures and syntax, see Quarantining Tests and the Quarantine Process handbook.

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 3.3 on GitLab.com, as well as for the default branch. To prepare for the next Ruby version, we run merge requests in Ruby 3.4. See the roadmap at Ruby 3.4 epic for more details.

To make sure all supported Ruby versions are working, we also run our test suite on dedicated 2-hourly scheduled pipelines for each supported versions.

For merge requests, you can add the following labels to run the respective Ruby version only:

• pipeline:run-in-ruby3_3

PostgreSQL versions testing

Our test suite runs against PostgreSQL 16 as GitLab.com runs on PostgreSQL 16 and Omnibus defaults to PG14 for new installs and upgrades.

We run our test suite against PostgreSQL 16, 17 and 18 on nightly scheduled pipelines.

[!note] With the addition of PG17, we are close to the limit of nightly jobs, with 1946 out of 2000 jobs per pipeline. Adding new job families could cause the nightly pipeline to fail.

Current versions testing

For the next Ruby versions we're testing against with, we run maintenance scheduled pipelines every 2 hours on the ruby-next branch. ruby-next must not have any changes. The branch is only there to run pipelines with another Ruby version in the scheduled maintenance pipelines.

The ruby-sync branch

The ruby-sync branch keeps the ruby-next and rails-next branches up-to-date with master. It is an orphan branch (not derived from master) that contains only its own .gitlab-ci.yml, a scripts/slack helper, and a README.md.

A scheduled pipeline runs on ruby-sync every 2 hours. The gitlab job:

1. Clones the full gitlab-org/gitlab repository.
2. For each of ruby-next and rails-next: checks out the branch, merges origin/master, and pushes the result.

No downstream pipelines are triggered by these pushes. The ruby-next and rails-next branches run their own scheduled maintenance pipelines independently.

Authentication

The gitlab job authenticates with a project token (RUBY_SYNC_TOKEN) that has write_repository scope and Maintainer role. The token is stored in the pipeline schedule variables for the ruby-sync branch.

Retry and failure notifications

The gitlab job retries once on script failure to handle transient errors. If it still fails, a notify job sends a message to the #backend Slack channel (username ruby-sync) through CI_SLACK_WEBHOOK_URL:

☠️ ruby-sync failed to merge master into the ruby-next/rails-next branches
in gitlab-org/gitlab. Pipeline: <pipeline_url> — Docs: <docs_url>

Troubleshoot ruby-sync failures

Common causes of failure:

• Transient GitLab load errors: The job clones the full repository, which can fail during high-load periods. Retry the failed pipeline.
• Merge conflicts: If ruby-next or rails-next has diverged from master in a way that causes conflicts, the git merge step fails. Manually resolve the conflict on the affected branch and retry.
• Authentication errors: Verify that the RUBY_SYNC_TOKEN project token has not expired. Check the pipeline schedule configuration.

To retry, go to the pipeline schedules page and run the ruby-sync schedule, or retry the failed job from the pipeline page.

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 7 on nightly scheduled pipelines, specifically when running forward-compatible PostgreSQL 15 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.

Single database tests run in two modes:

1. Single database with one connection. Where GitLab connects to all the tables using one connection pool. This runs through all the jobs that end with -single-db
2. Single database with two connections. Where GitLab connects to gitlab_main, gitlab_ci database tables using different database connections. This runs through all the jobs that end with -single-db-ci-connection.

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.

Elasticsearch and OpenSearch versions testing

Our test suite runs against Elasticsearch 9 as GitLab.com runs on Elasticsearch 9 when certain conditions are met.

We run our test suite against Elasticsearch 8, 9 and OpenSearch 1, 2 on nightly scheduled pipelines. All test suites use PostgreSQL 17 because there is no dependency between the database and search backend.

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.

CI configuration internals

See the dedicated CI configuration internals page.

Performance

See the dedicated CI configuration performance page.

Return to Development documentation