• @testing-library/react version: 16.3.0
• Testing Framework and version: Vitest 3.2.4
• DOM Environment: jsdom 26.1.0

Relevant code or config:

expect(
 await screen.findByText(
 "text which is supposed to be found after loading state is passed",
 ),
).toBeInTheDocument();

What you did:

Ran a test which failed on this assertion. Only sometimes. Typically in CI (Github Actions).

What happened:

×ばつ path/to/my.test.tsx > x > shows expected content 551ms
→ expect(element).toBeInTheDocument()
element could not be found in the document
⎯⎯⎯⎯⎯⎯⎯ Failed Tests 1 ⎯⎯⎯⎯⎯⎯⎯
FAIL path/to/my.test.tsx > x > shows expected content
Error: expect(element).toBeInTheDocument()
element could not be found in the document

Reproduction:

Does not reproduce easily. Is flaky. Trying to understand if there is some misunderstanding in how things are supposed to work.

Problem description:

When running a test based on findBy, we sometimes find them to be flaky. It is sometimes resolved by using waitForElementToBeRemoved on the previous state, but that just adds another possible race condition, and also tends to be flaky.

I expect the test to run for at least 1000ms (as per findBy logic documented here), but sometimes find that the test fails much sooner (after 551 ms in the example above).

Suggested solution:

I'm trying to deduce a cause here for further investigation.

• Is findBy resolving sooner than expected? If yes, it seems like a problem internal to the library, possibly Flaky tests: Asynchronous queries can return unmounted elements #865 ?
• Is the (vitest) reported time unreliable, and we may have hit the 1000ms timeout despite 551ms being reported? If so, maybe extending the default timeout could be beneficial, even though hundreds of similar tests tend to pass every time, so it seems odd.

Overall, this seems like a strange problem, and I'm not expecting a straight-forward fix, but would like you to know about it, and would be grateful for any pointers on how to analyze further.