Overview

A flaky test is a test that exhibit both a passing and failing status across multiple test runs for the same commit. If you commit some code and run it through CI, and a test fails, and you run it through CI again and the test passes, that test is unreliable as proof of quality code.

Flaky tests introduce risk and unpredictability into your CI system and end product. When people have to remember which tests are flaky, they lose trust in their test results, and a tremendous amount of time and resources are wasted on pipeline retries.

For each branch, the list shows the number of new flaky tests introduced by the commit, the number of flaky commits, total test time, and the branch’s latest commit details.

Use the following information to help prioritize flaky tests:

• Average duration: The average time the test takes to run.
• First flaked and Last flaked: The date and commit SHAs for when the test first and most recently exhibited flaky behavior.
• Commits flaked: The number of commits in which the test exhibited flaky behavior.
• Failure rate: The percentage of test runs that have failed for this test since it first flaked.
• Trend: A visualization that indicates whether a flaky test was fixed or it is still actively flaking.

Once you identify a flaky test you want to fix, click on the test to see links to view the most recent failed test run or the first flaky test run.

If a flaky test has not failed in the past 30 days, it is automatically removed from the table. You can also manually remove a flaky test by clicking on the trash icon that appears when you hover over the test row. It is added again if it re-exhibits flaky behavior.

Flaky tests in the default branch

The flaky test table for the default branch includes tests that have flaked in the default branch as well as any tests that have exhibited flakiness in a feature branch that was merged into the default branch.

Flaky tests from merged feature branches are found by checking which tests have exhibited flakiness in the most recent 5,000 commits using the Git commit history. The Git commit history is collected by the Test Visibility libraries and uploaded along with the test results every time the testing phase of a particular commit is executed in your CI build.

Limitations:

• If you squash or reset and force push commits in your feature branch, flaky tests that have been detected in that branch are not shown in the default branch because the commit history has been altered.
• If a flaky test is detected and subsequently fixed in the same feature branch, it still appears as a flaky test in the default branch, because the fix for the flaky test cannot be detected. However, you can manually remove that flaky test from the flaky tests table.

New flaky tests

New flaky tests are tests that exhibit flaky behavior and didn’t previously exist in the Flaky Tests table for the current branch or default branch of the repository.

Test Runs page

1. Navigate to the Test Runs page.
2. In the facets list on the left sidebar, expand the New Flaky facet in the Test section, and check true. All test runs that exhibited flaky behavior for the first time as per the definition above are displayed.

Branches page

1. On the Tests page, select the Branches view.
2. Filter the table to see branches, services, or commits of interest to you.
3. Look at the New Flaky column to see the number of new flaky tests introduced by the latest commit as per the definition above.

Ignore new flaky tests detected by mistake

You can ignore new flaky tests for a particular commit if you determine that those flaky tests were detected by mistake. The tests reappear if the commit exhibits flakiness again.

Click on the New Flaky number and then click Ignore flaky tests.

Known flaky failed tests

Known flaky failed tests are tests that have flaky behavior on the current or default branch of the repository.

Test Runs page

1. Navigate to the Test Runs page.
2. In the facets list on the left sidebar, expand the Known Flaky facet in the Test section, and check true. Failed test runs that were known to be flaky as per the definition above are displayed.

Branches page

1. On the Tests page, select the Branches view.
2. Filter the table to see any branches, services, or commits of interest.
3. The Failed column contains the number of failed tests and known flaky failed tests in the latest commit.

Further reading

Additional helpful documentation, links, and articles:

Learn about Test VisibilityDOCUMENTATION Flaky Tests OverviewKNOWLEDGE CENTER