Fix flaky sequential/test-domain-abort-on-uncaught and move back to parallel

  • Version: master
  • Platform: arm
  • Subsystem: test

See #11814 (comment)

This test is flaky under load. This can be seen by running it with a command like this:

tools/test.py -j 32 --repeat=32 test/parallel/test-domain-abort-on-uncaught.js

(If 32 isn't enough to cause problems on your setup, try 64 or higher.)

There are at least three solutions possible here:

  • move the test to sequential Done in test: fix flaky test-domain-abort-on-uncaught #11817
  • break the test up into as many as 13 separate test files so that each one is launching one additional node process rather than one test launching 13 of them in parallel
  • rewrite the test to run the 13 processes sequentially rather than in parallel

The second or third option above can be a good first contribution for someone.

How to fix

  • Read through the linked issue and understand the problem.
  • Replicate the failure on your machine.
  • Fix the test (probably using one of the methods suggested above).
  • Move the test back to test/parallel/ (it's currently in test/sequential/).
  • Follow CONTRIBUTING.md and submit a PR with the fix.

If you want to have a go at this comment on this issue to let us know. If you have any questions ask them here.