Run tests with gotestsum

[mhamza15]

Description

Switch tests to use gotestsum. gotestsum has a few benefits, namely prettier output, built-in junit report generation, automatic retries (individual tests, rather than full packages as some of our test infra does), and slow tests reporting.

The workflows will now report the top 20 slowest tests in the job summary. Click on any one of the jobs below to see the new output + job summary.

Backporting to improve release branches CI as well.

Related Issue(s)

Checklist

• "Backport to:" labels have been added if this change should be back-ported to release branches
• If this change is to be back-ported to previous releases, a justification is included in the PR description
• Tests were added or are not required
• Did the new or modified tests pass consistently locally and on CI?
• Documentation was added or is not required

Deployment Notes

AI Disclosure

Helped by Claude

[vitess-bot]

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

• Ensure that the Pull Request has a descriptive title.
• Ensure there is a link to an issue (except for internal cleanup and flaky test fixes), new features should have an RFC that documents use cases and test cases.

Tests

• Bug fixes should have at least one unit or end-to-end test, enhancement and new features should have a sufficient number of tests.

Documentation

• Apply the release notes (needs details) label if users need to know about this change.
• New features should be documented.
• There should be some code comments as to why things are implemented the way they are.
• There should be a comment at the top of each new or modified test to explain what the test does.

New flags

• Is this flag really necessary?
• Flag names must be clear and intuitive, use dashes (-), and have a clear help text.

If a workflow is added or modified:

• Each item in Jobs should be named in order to mark it as required.
• If the workflow needs to be marked as required, the maintainer team must be notified.

Backward compatibility

• Protobuf changes should be wire-compatible.
• Changes to _vt tables and RPCs need to be backward compatible.
• RPC changes should be compatible with vitess-operator
• If a flag is removed, then it should also be removed from vitess-operator and arewefastyet, if used there.
• vtctl command output order should be stable and awk-able.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 69.89%. Comparing base (32b8bd8) to head (0baa174).
⚠️ Report is 13 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #19076      +/-   ##
==========================================
- Coverage   69.89%   69.89%   -0.01%     
==========================================
  Files        1612     1613       +1     
  Lines      215826   216019     +193     
==========================================
+ Hits       150857   150989     +132     
- Misses      64969    65030      +61     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[arthurschreiber]

What do you think about adding the output of gotestsum tool slowest in a step after running the tests? Would help track down the biggest offenders to slow CI runs more systematically.

[mhamza15]

What do you think about adding the output of gotestsum tool slowest in a step after running the tests? Would help track down the biggest offenders to slow CI runs more systematically.

Good call!

[mhamza15]

Here's an example: https://github.com/vitessio/vitess/actions/runs/20826216650/attempts/1#summary-59828106838. Default is all tests >100ms. We have a gazillion, so changing to top 20.

[mattlord]

❤️

[arthurschreiber]

Only concern I have that this seems to stop CI from using vitessio/go-junit-report. I remember talking to @harshit-gangal some while back on why we have that fork and what we were using it for, but I don't quite remember what the reason was. I think it had something to do with flaky tests? 🤔

[arthurschreiber]

Digging a bit more, #9102 is when vitessio/go-junit-report was introduced.

[arthurschreiber]

And I think vitessio/go-junit-report@a07af90 was the main reason for the fork. For workflows that automatically re-run failed test cases, the text output across different runs was accumulated and then send to vitessio/go-junit-report, which then had to handle cases where a test might have failed but then was successful in a re-run.

Given that gotestsum does the re-running and report generation now, we might not need this anymore? Does that mean we can get rid of the remaining references to vitessio/go-junit-report in the code? test/ci_workflow_gen.go, .github/workflows/vitess_tester_vtgate.yml, .github/workflows/codeql_analysis.yml and a few more files still seem to have references to it.

[arthurschreiber]

According to gotestyourself/gotestsum#281, I think gotestsum does what we want. It records both the success and failure run, but the overall success of the test run is based on retries being eventually successful.

[arthurschreiber]

I also see there's different levels of retry logic. In config.json, a few test cases have RetryMax set to some specific value, with the default being 3. How does that interact with retrying at the gotestsum layer now (or the various test type bash scripts before)? Does this need to be cleaned up at some point?

[mhamza15]

@arthurschreiber Thank you for taking a deep look at go-junit-report ❤️ . I agree, I believe gotestsum does everything we need. We can even have it report the failures that were retried successfully so that we are also aware of flaky tests (the job will still pass). I'll clean up the other references I had missed.

As for retries, I intentionally left out retries for gotestsum for e2e tests (which uses config.json), and let test.go handle it. I first thought of passing RetryMax through to the gotestsum script and letting it handle it, but test.go also handles creating fresh VTDATAROOTs for each try, so I kept it simple for now for concern that retries would just fail immediately do to stale leftover state between runs.

I'm looking through it again and it looks like many tests themselves also create their own subdirectories under VTDATAROOT, so we might be able to remove that bit of logic, which allows us to rely on gotestsum's retries fully. I'll take a deeper look 👀 .

[mhamza15]

I've reverted the changes to use gotestsum for end-to-end tests since they require a bit more work to integrate nicely with the junit report generation. Will follow up in another PR.