CI: https://ci.nodejs.org/job/node-test-pull-request/37820/

You can see in #38226 where I bisected to determine that adding Proxy to bootstrap primordials is what caused this test to start failing much more frequently. I don't really understand the mechanics here, but maybe @not-an-aardvark or @aduh95 would be able to explain. (Or maybe not and it's a mystery.)

I still don't know if the issue here is with the test or with there really being a timing problem in Node.js core, but I'm guessing the latter but that it only manifests with a very fast CPU.

Regardless, fast-track to fix CI? Maybe leave the underlying issue open and investigate what's going on here?

(dqual-> equal in the commit message, but that can happen on landing.)

https://ci.nodejs.org/job/node-stress-single-test/293/ is a stress test to show improvement. It will probably still show some failures, but not nearly as many as https://ci.nodejs.org/job/node-stress-single-test/273/ from this morning which ran against master and failed 941 times out of 1000 runs.

@nodejs/testing

CI is green. Let's land this?

Stress test failed 192 times out of 1000 runs, which isn't great, but is way better than the current master branch which failed 941 times out of 1000 runs.

Would it be worth trying to identify which Proxy call (line 112 or line 124, or both) is actually causing the issue? It can happen in a follow up PR though.

Fast-track has been requested by @jasnell. Please 👍 to approve.

Landed in c7ccab3

Would it be worth trying to identify which Proxy call (line 112 or line 124, or both) is actually causing the issue? It can happen in a follow up PR though.

I've got another fix in the works that should make it possible to add back the Proxy() stuff removed here. Stay tuned.