ci: temporarily disable flaky VS Code extension E2E tests - #18413
Conversation
The VS Code extension E2E tests are flaky and currently failing in CI, blocking unrelated PRs. The zero-to-running shard intermittently fails in its "after each" teardown with an AggregateError on Windows (microsoft#18098), and because the extension_e2e_tests job gates CI, this fails runs on PRs that don't touch the extension. Temporarily disable the extension_e2e_tests job in tests.yml: - Force the job to always skip via `if: ${{ false && ... }}`, keeping the run_extension_e2e reference so the trigger-map binding test stays valid. - Remove the two aggregate-gate clauses (PR + non-PR) that treat a skipped e2e job as a failure, so the forced skip doesn't fail the overall gate. Extension unit tests (extension_tests_win, extension_bootstrap_linux) are unaffected and still run. Re-enabling is tracked by microsoft#18412. Refs microsoft#18098 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
|
🚀 Dogfood this PR with:
curl -fsSL https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.sh | bash -s -- 18413Or
iex "& { $(irm https://raw.githubusercontent.com/microsoft/aspire/main/eng/scripts/get-aspire-cli-pr.ps1) } 18413" |
There was a problem hiding this comment.
Pull request overview
This PR temporarily disables the flaky extension_e2e_tests job in .github/workflows/tests.yml. The zero-to-running E2E shard intermittently fails in its after each teardown with an AggregateError (tracked in #18098), and because the job gates CI, it was failing unrelated PRs. The job is force-skipped while keeping it wired into the workflow so the trigger-map binding test stays valid, and the aggregate-gate clauses that treat a skipped E2E job as a failure are removed. Re-enabling is tracked by #18412.
Changes:
- Force
extension_e2e_teststo always skip viaif: ${{ false && needs.setup_for_tests.outputs.run_extension_e2e == 'true' }}, retaining therun_extension_e2ereference so the binding test still passes. - Remove the PR and non-PR aggregate-gate clauses that fail CI when
extension_e2e_testsis skipped. - Update surrounding comments to explain the temporary disable and how to revert.
|
Retrying the failed CI jobs for this pull request from the CI run attempt. The rerun is being tracked in the rerun attempt. |
d619e31
into
microsoft:main
The extension E2E shards (specifically Windows, debug-dashboard) are consistently failing on main, blocking unrelated PRs from merging. Recent failures on main (last 5 days): - CI #25637 (Jun 28): Update OpenTelemetry.Instrumentation.StackExchangeRedis - CI #25621 (Jun 27): Prepare VS Code extension release v1.16.0 - CI #25616 (Jun 27): Fix VS Code extension CLI install - CI #25543 (Jun 26): Make dashboard launch opt-in by default Adds a `false &&` guard to the extension_e2e_tests job condition and removes the corresponding gate skip-checks, following the same pattern as PR #18413. Tracked by #18412.
The VS Code extension E2E tests are flaky and currently failing in CI, blocking unrelated PRs. The
zero-to-runningshard intermittently fails in itsafter eachteardown:(tracked in #18098). Because the
extension_e2e_testsjob gates CI, this flakiness fails runs on PRs that don't touch the extension.The fix
Temporarily disable the
extension_e2e_testsjob intests.yml:if: ${{ false && … }}(keeping therun_extension_e2ereference so the trigger-map binding test stays valid).Extension unit tests (
extension_tests_win,extension_bootstrap_linux) are unaffected and still run.Re-enabling
Tracked by #18412. Revert: drop the
false &&guard and restore the two gate clauses, after the flakiness in #18098 is fixed.Refs #18098