Hi, thanks for sharing this issue.

NUnit itself handles this under the hood (NCrunch doesn't call into the tests directly), but my understanding is that an async test should always be run to completion before the next one starts.

Note that this doesn't necessarily provide cover for other threads that are kicked off during the test run. If you have a test (async or otherwise) that launches another task and then returns before this task is complete, neither NCrunch or NUnit will know about the background thread in order to wait for it. This thread can then freely interfere with subsequent test runs in the process.

If adjusting your code to cater for such a scenario is too difficult (i.e. you can't get access to the Task management from the test code), using the IsolatedAttribute would probably solve this for you as NCrunch would just clean up the process and avoid re-using it. There will be some loss of performance involved in re-spawning the process after the test completes.