We determined this failure is an environment state issue rather than a code regression. The error originates from a stale compiled dist artifact (e2e/e2e-utils/dist/esm/e2eSetupTeardown.js) that was missing the e2eStartDummyServer export at test execution time, even though the source file correctly defines it and the PR makes no changes to e2e-utils. The dist has since been rebuilt correctly, confirming this was a build ordering issue in the CI environment.
Failed after FFloView latest
Always execute beforeLoad for client lanes (#7922)
78051d20a9dOpen on GitHubSelf-Healing CI
Classification: Environment state
An empty commit was applied to the branch to trigger a new CI pipeline execution to resolve the environment issue.
The failure is not a code change for the following reasons:
The PR does not touch e2e-utils source files. None of the files in
e2e/e2e-utils/appear in the PR diff. The source filee2eSetupTeardown.tscorrectly exportse2eStartDummyServerand was not modified by this PR.The error is in a compiled build artifact, not source. The error occurs in
e2e/e2e-utils/dist/esm/e2eSetupTeardown.js(a compiled output file). At the time the test ran, this dist file was stale and did not include the export. The current dist file on disk is correct, confirming the source was never wrong — only the build artifact was outdated.The dist was rebuilt after the test failure. The fact that the dist now contains the correct exports, while the test failed with a missing export, indicates the e2e-utils package was not rebuilt before the test executed. This is a build ordering/environment issue, not a code regression.
Flakiness rate is 0% and no similar failures in other branches. The task has never been flaky, and the same failure does not appear in other branches, ruling out both a flaky test and a pre-existing systemic regression.
Conclusion: The failure is purely an environment state issue — the e2e-utils dist was stale when the CI job ran the Playwright tests.