We did not make any code changes, because we could not find a code cause for the e2e config.webServer timeouts. Locally, the mock API starts under tsx 4.23.15 and the davinci app builds and serves on port 5829. The 60s timeout in CI is more likely the cold build && vite start-up than the tsx bump, so these tasks may only need a rerun or a longer webServer timeout.
Failed after Rrenovate[bot]
chore(deps): update dependency tsx to v4.23.15
8585c1badeOpen on GitHubSelf-Healing CI
Classification: Code change needed
Fix generated
Verification not started
No files were changed, and no commit message applies.
What failed: The three e2e-ci tasks all failed with "Timed out waiting 60000ms from config.webServer". The PR only bumps tsx from 4.21.0 to 4.23.15 in pnpm-workspace.yaml.
What was checked:
.nx/SELF_HEALING.mddoes not exist, and AGENTS.md and CLAUDE.md have no predefined fixes or off-limits areas.- pnpm-lock.yaml already pins tsx 4.23.15, and the installed binary reports v4.23.15, so the version bump is consistent.
pnpm --filter am-mock-api exec tsx src/index.jsstarts and listens on port 9443 under the new tsx.pnpm --filter @forgerock/davinci-app... buildsucceeds andviteserves on port 5829. The build was served from the Nx cache, so a cold build was not tested.- The branch 857 comparison found no matching failure, so the error is not pre-existing there.
Why nothing was changed:
- The failure did not reproduce locally.
- The CI output shows only the timeout message, with no stack trace or startup log.
- The most likely explanation is that a cold CI
build && vitestart takes longer than the 60s webServer timeout. This is unconfirmed. - Raising the timeout, skipping tests, or pinning tsx back to 4.21.0 would not be justified by this evidence.
- The failing tasks need their web server startup logs, or a rerun, to find the actual cause.