test: work around ENOTEMPTY when cleaning tmp dir

[bnoordhuis]

Replace the homegrown rimrafsync implementation in test/common with
a call to fs.rmdirSync(path, { recursive: true }).

Fixes: #30620
Fixes: #30844

cc @joaocgreis

[cjihrig]

I think this will fail without the first commit from #30569 due to encodings. And without #30785 it seems (based on CI runs) that Windows will be flaky.

[bnoordhuis]

It's currently failing like this so that sounds plausible.

Error: EPERM: operation not permitted, unlink '\\?\C:\workspace\node-test-binary-windows-js-suites\node\test\.tmp.63\node-copy.exe'
    at unlinkSync (fs.js:1066:3)
    at fixWinEPERMSync (internal/fs/rimraf.js:245:5)

[bnoordhuis]

Still the same error after rebasing on top of master (now that #30785 is merged):

05:53:31     Error: EPERM: operation not permitted, unlink '\\?\C:\workspace\node-test-binary-windows-js-suites\node\test\.tmp.63\node-copy.exe'
05:53:31         at unlinkSync (fs.js:1066:3)
05:53:31         at fixWinEPERMSync (internal/fs/rimraf.js:256:5)

[cjihrig]

Looking at the full stack trace, I think I see what the problem is. In #30785, I updated the synchronous retry logic where the original rimraf code just retried in a loop. But it's possible that that code is never even reached if errors are encountered in the readdirSync() + rimrafSync() loop a few lines up.

[bcoe]

Is there a chance we also run into a race condition on unlinkSync; it looks like we retry with backoff on rmdirSync (for directories) but not for unlinkSync for files.

Seems like using similar logic for both would be appropriate?

[cjihrig]

@bcoe yep, that's exactly what I did in the most recent commit in #30569. On one CI run, it appears to have worked. I'll rerun the CI a few times to verify.

[cjihrig]

@bnoordhuis if you rebase this now things should, hopefully, be working 🤞

[cjihrig]

Can you also remove the options from refresh() in test/common/README.md? See https://github.com/nodejs/node/pull/30888/files#diff-6057afb393d4ecf6a2cc0937eec56877L901

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/27586/

[bnoordhuis]

Thanks, Colin. Updated + rebased: https://ci.nodejs.org/job/node-test-pull-request/27586/

[Trott]

Are the changes to test/parallel/parallel.status intentional or unrelated?

[Trott]

LGTM pending parallel.status question.

[Trott]

Are the changes to test/parallel/parallel.status intentional or unrelated?

Oh, I see, yeah, it's intentional. Ignore my question.

[Trott]

Landed in edf654d