ci: stabilize TMPDIR for Go test cache

[davdhacs]

Description

Set TMPDIR=/tmp in the unit-tests workflow env block.

Go's test cache (computeTestInputsID) hashes every environment variable that the test binary reads via os.Getenv(). These reads are recorded in the test binary's internal log (via testing/internal/testdeps) and replayed by computeTestInputsID on subsequent runs to verify that inputs haven't changed.

Approximately half of our test packages call os.TempDir() (directly or through imported libraries), which reads TMPDIR. On GHA, each job gets a unique TMPDIR — so testInputsID differs between the master push (cache save) and PR run (cache restore), causing Phase 2 cache misses for every affected package.

Config	Cache hits
Master baseline (no fix)	571/762 (75%)
With TMPDIR=/tmp (from master cache)	693/762 (91%)
With TMPDIR=/tmp (own cache, stable)	712/762 (93%)

GODEBUG=gocachehash=1 on CI (PR #21276) captured 128,527 testInputs hash lines across 762 packages. 817 of those reference TMPDIR.

Testing and quality

• CI results are inspected

Automated testing

No test changes. This fixes the CI test caching infrastructure.

How I validated my change

• GODEBUG=gocachehash=1 on CI captured hash inputs confirming TMPDIR in testInputsID (PR debug(ci): testInputsID hash capture for cache investigation #21276, run 27789436790)
• 10-branch parallel A/B experiment (PRs test(ci): cache-test-a — baseline: v2 trim + coverprofile #21199-test(ci): cache-test-j — coverprofile, NO modfix (control) #21215) isolated variables
• PR ci: stabilize TMPDIR for Go test cache #21277 (experiment version) cherry-picked 11 master commits demonstrating 93% cache hits across real code changes vs 75% baseline
• Local reproduction confirmed TMPDIR is not hashed when stable

[openshift-ci]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Enterprise

Run ID: a7248273-a048-4b76-9fee-f2bb12634712

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch davdhacs/tmpdir-fix

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

🚀 Build Images Ready

Images are ready for commit 3f08c9a. To use with deploy scripts:

export MAIN_IMAGE_TAG=4.12.x-239-g3f08c9a93c

[tommartensen]

What is a "Phase 2 cache" in this context?

[davdhacs]

What is a "Phase 2 cache" in this context?

Claude named the key creation steps "Phase 1" and "Phase 2" while I was working with it. The first phase is a content hash key lookup, and the second includes the env vars and causes our miss because the TMPDIR changes every run in GHA's runner (and some tests read the TMPDIR variable):

https://go.googlesource.com/go/+/refs/heads/master/src/cmd/go/internal/test/test.go#1882

// The test cache result fetch is a two-level lookup.
--
//
// First, we use the content hash of the test binary
// and its command-line arguments to find the
// list of environment variables and files consulted
// the last time the test was run with those arguments.
// (To avoid unnecessary links, we store this entry
// under two hashes: id1 uses the linker inputs as a
// proxy for the test binary, and id2 uses the actual
// test binary. If the linker inputs are unchanged,
// this way we avoid the link step, even though we
// do not cache link outputs.)
//
// Second, we compute a hash of the values of the
// environment variables and the content of the files
// listed in the log from the previous run.
// Then we look up test output using a combination of
// the hash from the first part (testID) and the hash of the
// test inputs (testInputsID).

[rhacs-bot]

/retest

[davdhacs]

I am seeing 93% go unittest cache hits on recent master commits without this change. I need to investigate more to verify if this change is needed.

[rhacs-bot]

/retest