[FLINK-36646] Test different versions of the JDK in the Flink image

[SamBarker]

What is the purpose of the change

• adding tests to demonstrate FLINK-36646

Brief change log

• adding tests to demonstrate FLINK-36646

Verifying this change

This change is already covered by existing tests, such as test_application_operations.sh should pass with each of the supported JDK versions.

Does this pull request potentially affect one of the following parts:

• Dependencies (does it add or upgrade a dependency): no
• The public API, i.e., is any changes to the CustomResourceDescriptors: no
• Core observer or reconciler logic that is regularly executed: no

Documentation

• Does this pull request introduce a new feature? no
• If yes, how is the feature documented? not applicable

[SamBarker]

cc: @tomncooper & @robobario

[robobario]

I'm wondering if it's a bit funky that we are using java-version to control both the JDK/JRE used to build/run the operator and the runtime of the flink image. Aren't they different dimensions of the matrix?

This would show that a JDK17 operator works with a JDK17 flink image, but not cover the current default operator runtime with a JDK17 flink image.

[SamBarker]

Yeah, I had debated making a flink-tag parameter but didn't as we would then have to manually encode the matrix.

This would show that a JDK17 operator works with a JDK17 flink image, but not cover the current default operator runtime with a JDK17 flink image.

I'm not convinced that the operator JDK really matters so don't think its particularly important to have a JDK11 operator deploying a JDK 17 Flink. I can go that direction if others see it as valuable.

[robobario]

yeah you're right we aren't likely to see some incompatibility arising from different JRE combinations. Maybe it would be better to hold the operator JDK constant and vary only the flink runtime. Just to make it clearer which thing is varied.

[tomncooper]

Yeah I agree with @SamBarker. The issue here is actually not the interplay of JDK versions but rather the config supplied by the operator's logic not being compatible (or more accuratly enabling) Java 17 support. So that's what we need to test.

[robobario]

maybe this should be an optional flink-java-version and the matrix could contain the versions we want to test

[tomncooper]

LGTM, just need to update the changelog in the PR description.

[SamBarker]

E2E test failed as expected: https://github.com/apache/flink-kubernetes-operator/actions/runs/11696768142/job/32793460638?pr=910#step:9:4256

[gyfora]

Closing this as it was merged in another PR