fitzee commented on PR #43944:
URL: https://github.com/apache/superset/pull/43944#issuecomment-5564426583

   ### Investigation of the shard 3 submenu failure
   
   Investigated the [failed upstream 
job](https://github.com/apache/superset/actions/runs/34074363889/job/101598080153)
 on unchanged head 7700bbb3e35cbd37a41f9605de4cc2069ef5ad13. No code/test edits 
or new commits are justified by the evidence so far.
   
   - Installed frontend dependencies with npm ci under Node v24.16.0. Actual 
user-event is **12.8.3**: its hover implementation synchronously fires 
pointer/mouse events, so adding await would not fix this. antd is 6.6.2, 
@rc-component/menu 1.5.0, and Jest 30.5.0.
   - The menu always includes Export All Data; feature/permission branches 
affect other items or disabled exports, not that submenu's existence. The 
submenu uses a 100 ms opening delay followed by a requestAnimationFrame 
visibility update; Testing Library's async query default is 1 second. This is a 
possible timing surface, **not an established root cause** from the truncated 
CI DOM.
   - Compared with base 87f138cc80ac9de3a3fcee082c41ba2368be8186: 
Explore/header/menu code, test setup, dependency manifests/lockfile, and 
user-event version are unchanged. The PR's bootstrap type change is type-only; 
its new runtime diagnostics are in the embedded entry path. This comparison 
alone does not prove flakiness.
   
   Actual local results:
   - Isolated reported test: **1 passed**.
   - Complete ExploreChartHeader suite on PR: **41 passed**; same suite on an 
archived base using the same installed dependencies: **41 passed** initially 
and in two subsequent runs.
   - Controlled PR shard 3 with coverage: **161 suites / 1,427 tests passed**, 
including ExploreChartHeader (14.518 seconds).
   - Focused embedded diagnostics/index tests: **2 suites / 18 tests passed**.
   - ExploreChartHeader with --runInBand --detectOpenHandles: **41 passed**, 
exit 0, no open-handle report.
   
   Commands from superset-frontend, with Node v24.16.0 on PATH and CI=true, 
NODE_ENV=test, NODE_OPTIONS=--max-old-space-size=8192:
   
       node_modules/.bin/jest --silent --coverage --shard=3/8 
--coverageReporters=json --maxWorkers=3
       node_modules/.bin/jest --silent 
src/explore/components/ExploreChartHeader/ExploreChartHeader.test.tsx 
--runInBand --detectOpenHandles
       node_modules/.bin/jest --silent 
src/embedded/guestTokenDiagnostics.test.ts src/embedded/index.test.tsx 
--maxWorkers=2
   
   One exploratory over-parallel invocation was discarded: combining the npm 
script's --max-workers option with an appended --maxWorkers option resolved to 
80 workers, causing unrelated local timeouts. The controlled shard and focused 
reruns invoke Jest directly with a single worker setting.
   
   **Teardown warning is separate:** the force-exit warning also appeared on 
the fully passing local shard. It therefore does not establish the cause of 
this submenu failure; the leaking worker/suite has not been identified. No 
warning suppression or timer cleanup changes are proposed.
   
   **Rerun needed:** requested one rerun of failed job 101598080153 via 
GitHub's job-rerun endpoint; GitHub returned HTTP 403. No rerun started 
(workflow remains completed/failure, attempt 1). A maintainer with Actions 
permissions should rerun that failed job once. This remains an unreproduced 
failure with a suspected timing cause, not a proven pre-existing flake or a 
fixed CI failure.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to