[SPARK-54889][INFRA][FOLLOWUP] Enable codecov test report on all stable workflows

[gaogaotiantian]

What changes were proposed in this pull request?

Enable codecov test report upload on all workflows in apache/spark, which are all scheduled tests plus main commit.

Why are the changes needed?

After experimenting with codecov, we had a better understanding of how things work. They use flag to categorize tests from different environments. This is helpful for us to do test analysis. Our schedule tests are sometimes unstable, this could help us to find the unstable tests.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

This is a CI change so we had to wait for CI.

Was this patch authored or co-authored using generative AI tooling?

No

[github-actions]

JIRA Issue Information

=== Test SPARK-54889 ===
Summary: Upload test result to codecov.io to have a test dashboard
Assignee: Tian Gao
Status: Resolved
Affected: ["4.2.0"]

---

This comment was automatically generated by GitHub Actions

[gaogaotiantian]

@zhengruifeng the CI actually passed after re-run (some glitches). Don't know why github did not show it correctly.

[zhengruifeng]

I remember workflow with codecov is much slower, I am not sure whether we want to enable it on all workflows

[HyukjinKwon]

yeah we can't it becomes much much slower and unstable ..

[gaogaotiantian]

I believe you are thinking about "coverage run", where we instrument the unittests so we can generate coverage data. This is a different thing. We do not change how we run tests at all. The only thing we do is to upload already generated test report (not coverage report) to codecov.io.

Codecov.io is the website to visualize data. It is often used to visualize coverage data - which is generated by coverage runs (we do that daily). In this case, we are using a different feature to visualize test results, it's not related to coverage at all.

[HyukjinKwon]

In this case, we are using a different feature to visualize test results, it's not related to coverage at all.

What are we uploading?

[HyukjinKwon]

Do you have any example to show? how would it look like?

[gaogaotiantian]

We are uploading the xml files generated by each unittest. Notice we are already uploading those to github artifacts in another action. The dashboard would look like https://app.codecov.io/gh/apache/spark/tests - it's not the most delicate dashboard, but it's something. The cost is really low and we at least have some test analysis (I need to work on it a bit more to have better data, but this is a start)