Not sure why a bump of Jacoco would start failing these tests. It might be that they have always been flaky and that they rely on ordering. Do you mind debugging locally what is going on here and if you can reproduce the same issues with the same Jacoco version, but with different ordering?

Codecov Report

Attention: 2 lines in your changes are missing coverage. Please review.

Comparison is base (fd62884) 85.52% compared to head (d74808b) 85.50%.

❗ Current head d74808b differs from pull request most recent head 56b47c0. Consider uploading reports for the commit 56b47c0 to get more accurate results

@@             Coverage Diff              @@
##               main    #3147      +/-   ##
============================================
- Coverage     85.52%   85.50%   -0.02%     
  Complexity     2912     2912              
============================================
  Files           333      333              
  Lines          8856     8858       +2     
  Branches       1095     1096       +1     
============================================
  Hits           7574     7574              
- Misses          994      995       +1     
- Partials        288      289       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

@TimvdLippe the root cause of failures is this change in JaCoCo: jacoco/jacoco@b865890. org.mockito.internal.util.reflection.ReflectionMemberAccessor#silentSetAccessible doesn't work without JVM flags opening packages (on Java 17).

So my question is: what should strategy be around this problem? is it ok to add flags for unit tests or is there any other global solution required?

@raphw I think adding these flags for our tests is okay, but what do you think?

I think so, yes. I'd go for it.

CI is green, one job fails at Codecov upload:

Error: Codecov: Failed to properly upload: The process '/home/runner/work/_actions/codecov/codecov-action/v3/dist/codecov' failed with exit code 255