Investigate flaky test-fs-watch-encoding

[joaocgreis]

• Version: unrelated change rebased on master
• Platform: osx1010
• Subsystem: test

https://ci.nodejs.org/job/node-test-commit-osx/3726/nodes=osx1010/tapTestReport/test.tap-598/

not ok 598 parallel/test-fs-watch-encoding
# TIMEOUT

Test source: https://github.com/nodejs/node/blob/master/test/parallel/test-fs-watch-encoding.js

cc @nodejs/testing

[santigimeno]

It's the same issue as: #4629.

[Trott]

Stress test results are surprising. Maybe I'm doing something wrong?: https://ci.nodejs.org/job/node-stress-single-test/755/nodes=osx1010/console

[Trott]

OK, looks like the test just isn't calling common.refreshTmpDir(). That's the issue with the stress test results. But it's still flaky and needs a timer on OS X too. PR coming in a few minutes....

[Trott]

Stress test with tmp dir being refreshed: https://ci.nodejs.org/job/node-stress-single-test/756/nodes=osx1010/console

[Trott]

Hmmm, no failure in that stress test with the refreshTmpDir() added. That surprises me. Let me try upping the number of runs from 999 to 9999. If it still doesn't fail, maybe declare it fixed, at least until we see it again: https://ci.nodejs.org/job/node-stress-single-test/757/nodes=osx1010/console

[santigimeno]

FWIW I can consistently reproduce the issue locally even after adding refreshTmpDir(). OS X version 10.11.5

[Trott]

@santigimeno How are you running it? (Via tools.py or directly with node? Current master or v6.2.1 or something else? Anything else that might be relevant...)

I'm on 10.10.5 and unable to reproduce, but maybe I'm just not running it enough times or something?

[santigimeno]

@Trott Current master. I can easily reproduce it using this script nodejs/testing#3 (comment). Also, running the test suite non-stop for a while reproduces the issue after a while (1 hour or so)

[Trott]

Is this still a problem now that #7356 landed 9 days ago?

[Trott]

I'm unable to replicate this problem now since 5267f29 landed. If I revert 5267f29, then I can easily replicate this problem.

So...I'm going to close this. Of course, feel free to re-open and/or comment if I am in error.