test(fetch): make prepareData async to enforce fakeTime order

[Ugzuzg]

Which problem is this PR solving?

Trying to resolve flaky fetch instrumentation tests.

Fixes #2852

Short description of the changes

I was able to reproduce the issue by supplying half a cpu to a docker container in which the tests are running.

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

Repeated the test execution multiple times in a container with half a cpu.

Checklist:

• Followed the style guidelines of this project
• Unit tests have been added
• Documentation has been updated

[codecov]

Codecov Report

Merging #2969 (d4e3750) into main (e44f6d1) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main    #2969   +/-   ##
=======================================
  Coverage   92.77%   92.77%           
=======================================
  Files         183      183           
  Lines        6134     6134           
  Branches     1300     1300           
=======================================
  Hits         5691     5691           
  Misses        443      443           

[dyladan]

Looks like coverage changes imply the browser tests no longer trigger the isBrowser check in the browser detector. I'm not sure what in the diff would have caused that

[Ugzuzg]

@dyladan, seems to be an issue with coverage reporting. The reports are uploaded under different commit hashes for some reason. Both runs belong to Unit Tests #4251:

• regular unit tests https://github.com/open-telemetry/opentelemetry-js/runs/6425366704?check_suite_focus=true#step:8:43

  View report at: https://codecov.io/github/open-telemetry/opentelemetry-js/commit/a5075e8a9276625dc62418c2b9c803d67ba49b94
  

• browser unit tests https://github.com/open-telemetry/opentelemetry-js/runs/6425366115?check_suite_focus=true#step:9:43

  View report at: https://codecov.io/github/open-telemetry/opentelemetry-js/commit/0484412413f60e10ac9a007fd573ca393d196087
  

Looks better after rerunning, but it is indeed strange. The same thing happened in my other PR, I think.

[rauno56]

Any pointers on how to repro, @Ugzuzg ?

[Ugzuzg]

@rauno56, I'm mounting this repo in a container with half a cpu: docker run --cpu 0.5 -v $(pwd):/app -it circleci/node:16-browsers /bin/bash. Then building and running the tests of fetch instrumentation. They're failing on almost every execution on main.

[dyladan]

LGTM thanks for looking into this

[dyladan]

@rauno56 the race you mentioned is removed. Would appreciate a ✅ and we can merge

[rauno56]

I really appreciate the effort you've put into it finding out the root cause, @Ugzuzg. Thanks!

[pichlermarc]

Looks great! Thanks for putting in the time and effort to work on this. 🙂

[legendecas]

Thank you for your work on this!

[legendecas]

Landed. Thank you for your contribution!