[christianaurichzm]

test-fastutf8stream-periodicflush failed twice in the 2026-11-09 reliability report with EBADF on write.

The manual flush case calls flush() and then destroy() right away. destroy() closes the fd without waiting for the write that flush() started, so on a busy machine the close can happen first and the write fails. The file was also read on a 500 ms timer instead of after the flush.

This moves the destroy() and the read into the flush callback.

It doesn't fail for me locally, but delaying every fs.write() by 50 ms with the preload below gives the same stack trace as CI. With it the test fails 196/200 on main and 0/200 with this change.

const fs = require('fs');
const orig = fs.write;
fs.write = function(...args) { setTimeout(() => orig.apply(fs, args), 50); };

Refs: https://github.com/nodejs/reliability/blob/main/reports/2026-11-09.md

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.69%. Comparing base (768315c) to head (36322b3).
⚠️ Report is 21 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #67355      +/-   ##
==========================================
- Coverage   90.69%   90.69%   -0.01%     
==========================================
  Files         797      797              
  Lines      279572   279572              
  Branches    53778    53788      +10     
==========================================
- Hits       253569   253565       -4     
- Misses      16287    16297      +10     
+ Partials     9716     9710       -6     

see 29 files with indirect coverage changes

🚀 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.