Fix verification of version probing

[cadonna]

Committer Checklist (excluded from commit message)

• Verify design and implementation
• Verify test coverage and CI build status
• Verify documentation (including upgrade notes)

[mjsax]

When did we bump the number?

[cadonna]

Here: #10609

[showuon]

This manually number bumping is error-prone. Might need a better way to handle it in the future. What do you think?

[cadonna]

@showuon Agreed! All of us stumbled upon this. To be fair, usually we forget to adapt the test, but this time the author remembered to modify it, but missed to change one number. Such a situation could be changed by using variables. Let me have a stab at this.

[ableegoldman]

Wait, I'm confused. I definitely bumped the version number in this test in that PR you referenced (scroll down to the end of the changes, it's the last file). Did we for some reason bump the protocol version number again since then?

[ableegoldman]

Oh nevermind, I see, I just missed bumping one instance of it in the message. In that case this LGTM

[ableegoldman]

Thanks for cleaning this up to use a variable and hopefully reduce how many times we have to fix this test going forward

[chia7712]

LGTM

[chia7712]

Could you add comments about why/when we need to bump this value? That can helps us from debugging this issue (again) in the future.

[ableegoldman]

Oh I actually added a comment in StreamsAssignmentProtocolVersions that describes everything you need to do when bumping the protocol number, and when you should/shouldn't bump it. It should cover everything but now the 1st instruction should be updated to say "bump the highest_version variable" instead of saying to update these messages directly @cadonna

[cadonna]

Good idea! I updated the instructions.

[cadonna]

I verified this fix:
http://confluent-kafka-branch-builder-system-test-results.s3-us-west-2.amazonaws.com/2021-07-12--001.1626105849--cadonna--fix_version_probing_system_test--9f19bb236/report.html

[cadonna]

Failed tests are unrelated and known to be flaky.

[cadonna]

@kkonstantine Cherry-picked to 3.0 since it fixes a consistently failing system test also there.

[cadonna]

Here the system test result on 3.0:
http://confluent-kafka-branch-builder-system-test-results.s3-us-west-2.amazonaws.com/2021-07-12--001.1626111082--cadonna--fix_version_probing_system_test_3.0--659f56ce5/report.html