NodeJs fatal error when running tests

[Demivan]

Describe the bug

When running tests on Windows I get a core dump from Node.

It is not consistent. Sometimes it works: https://github.com/fluent-vue/rollup-plugin-fluent-vue/runs/4635574629?check_suite_focus=true#step:6:17 and sometimes not: https://github.com/fluent-vue/rollup-plugin-fluent-vue/runs/4635579648?check_suite_focus=true

I don't think it is the same as #317. I tried to get stack trace for that issue to compare but it didn't work.

Reproduction

https://github.com/fluent-vue/rollup-plugin-fluent-vue

System Info

System:
    OS: Windows 10 10.0.17763
    CPU: (2) x64 Intel(R) Xeon(R) CPU E5-2673 v4 @ 2.30GHz
    Memory: 4.99 GB / 7.00 GB
  Binaries:
    Node: 16.13.1 - C:\hostedtoolcache\windows\node\16.13.1\x64\node.EXE
    Yarn: 1.22.17 - C:\npm\prefix\yarn.CMD
    npm: 8.1.2 - C:\hostedtoolcache\windows\node\16.13.1\x64\npm.CMD
  Browsers:
    Internet Explorer: 11.0.17763.2145
  npmPackages:
    vite: ^2.7.7 => 2.7.7 
    vitest: ^0.0.114 => 0.0.114 

Used Package Manager

pnpm

Logs

https://github.com/fluent-vue/rollup-plugin-fluent-vue/runs/4635579648?check_suite_focus=true


FATAL ERROR: v8::ToLocalChecked Empty MaybeLocal.

 1: 00007FF6417E30AF v8::internal::CodeObjectRegistry::~CodeObjectRegistry+112511
 2: 00007FF641772216 DSA_meth_get_flags+65542
 3: 00007FF6417730CD node::OnFatalError+301
 4: 00007FF6420AD5D5 v8::V8::ToLocalEmpty+53
 5: 00007FF64176A2B6 DSA_meth_get_flags+32934
 6: 00007FF64175DC1B v8::base::CPU::has_sse+42267
 7: 00007FF641845BF7 uv_timer_stop+1207
 8: 00007FF6418421CB uv_async_send+331
 9: 00007FF64184195C uv_loop_init+1292
10: 00007FF641841AFA uv_run+202
11: 00007FF641810C05 node::SpinEventLoop+309
12: 00007FF6416AB4D0 v8::internal::wasm::SignatureMap::Freeze+35776
13: 00007FF6416A6B28 v8::internal::wasm::SignatureMap::Freeze+16920
14: 00007FF6418322DD uv_poll_stop+557
15: 00007FF642643DF0 v8::internal::compiler::RepresentationChanger::Uint32OverflowOperatorFor+146416
16: 00007FFB0E117974 BaseThreadInitThunk+20
17: 00007FFB10DEA2F1 RtlUserThreadStart+33

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines.
• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Check that this is a concrete bug. For Q&A open a GitHub Discussion or join our Discord Chat Server.
• The provided reproduction is a minimal reproducible example of the bug.

[Demivan]

And I get similar issue on linux too. But with different project (will sent a link when I'm done with the branch)

Like with Windows all tests pass but after that node crashes.

  System:
    OS: Linux 5.15 Arch Linux
    CPU: (16) x64 AMD Ryzen 7 3700X 8-Core Processor
    Memory: 2.39 GB / 15.60 GB
    Container: Yes
    Shell: 3.3.1 - /bin/fish
  Binaries:
    Node: 17.3.0 - /usr/bin/node
    Yarn: 1.22.17 - /usr/bin/yarn
    npm: 8.3.0 - /usr/bin/npm
  Browsers:
    Chromium: 96.0.4664.110
    Firefox: 95.0.2
  npmPackages:
    vite: ^2.7.7 => 2.7.7 
    vitest: ^0.0.114 => 0.0.114 

FATAL ERROR: v8::FromJust Maybe value is Nothing.
 1: 0x55cc11bca6b1 node::Abort() [node]
 2: 0x55cc11ad0ceb node::FatalError(char const*, char const*) [node]
 3: 0x55cc11daa3fb v8::Utils::ReportApiFailure(char const*, char const*) [node]
 4: 0x55cc11bcf0fe node::fs::FileHandle::CloseReq::Resolve() [node]
 5: 0x55cc11bcfa5a  [node]
 6: 0x7f22568dc302  [/usr/lib/libuv.so.1]
 7: 0x7f22568dffcd  [/usr/lib/libuv.so.1]
 8: 0x7f22568f4266  [/usr/lib/libuv.so.1]
 9: 0x7f22568e0897 uv_run [/usr/lib/libuv.so.1]
10: 0x55cc11afe1b7 node::SpinEventLoop(node::Environment*) [node]
11: 0x55cc11c9f98c node::worker::Worker::Run() [node]
12: 0x55cc11c9fe18  [node]
13: 0x7f2255bb1259  [/usr/lib/libpthread.so.0]
14: 0x7f2255ada5e3 clone [/usr/lib/libc.so.6]

[antfu]

Yeah, I got that sometimes as well. Not sure how to debug it. Help wanted.

[userquin]

I think it is something related with pnpm/shell/npx, on windows all tests pass individually, but running nr ci or pnpm run test:ui or pnpm run test:all on the repo fails a number of times, the error switchs from the windows cmd shell, to npx to node, it is weird.

Take a look at the video on discussions room on vitest repo (50s of video running test:all a few times until the test pass again). If I change something, then nr ci works first time, then will fail again (if nothing changes).

[Demivan]

vitest/packages/vitest/src/node/pool.ts

Line 99 in 801d939

close: () => pool.destroy(),

Destroying the worker pool is causing the crash. Not sure why yet.

[antfu]

vitest/packages/vitest/src/node/pool.ts

Line 99 in 801d939

close: () => pool.destroy(),

Destroying the worker pool is causing the crash. Not sure why yet.

Oh really? It does make the error not appearing at least for me, Thanks! I think we don't actually have to have that line. Just commented it out and let's see how it goes