fix: allow pushing images with different arch/os to docker daemon

[mpeddada1]

Fixes #4265

• Avoid throwing exception when building image with mismatched arch/os to the local docker daemon.
• Follow convention of using event handlers to log warning when building for matching os/arch. Example: https://github.com/GoogleContainerTools/jib/blob/b06b001ed26b6a1fedc1ac738f628513390433e9/jib-core/src/main/java/com/google/cloud/tools/jib/builder/steps/PullBaseImageStep.java#L225

[mpeddada1]

Notes on kokoro-ubuntu timeout:

• The job passed on https://github.com/GoogleContainerTools/jib/pull/4268/commits/f2d886b289a2e0d471f37281ce23322e00e6a288 but started failing after test for arm64 was added.

Unable to reproduce locally with ./gradlew clean build :jib-core:integrationTest.

kokoro-macos:

• The job is failing during job start up:

  ++ docker-machine ls
  NAME   ACTIVE   DRIVER   STATE   URL   SWARM   DOCKER   ERRORS
  ++ docker-machine start default
  Docker machine "default" does not exist. Use "docker-machine ls" to list machines. Use "docker-machine create" to add a new one.

[chanseokoh]

@mpeddada1 just so you know, it sounds like #4267 is another issue introduced by #4249 in 3.4.3.

[mpeddada1]

@mpeddada1 just so you know, it sounds like #4267 is another issue introduced by #4249 in 3.4.3.

Thank you @chanseokoh! Given that it is not reproducible in 3.4.2, the change in logging behavior may connected to some of the modifications made in https://github.com/GoogleContainerTools/jib/pull/4249. Will take a deeper look to understand what causes the difference in behavior for windows (perhaps in a separate dedicated PR).

[mpeddada1]

Regarding the integration test behavior on the CI:

• Changing test names is causing different behavior on the ubuntu CI. Documenting test names to better understand execution order: https://github.com/GoogleContainerTools/jib/pull/4268/commits/d92123d9675a3ea662c8c1a77e757bf003ffe0c6. According to https://github.com/junit-team/junit4/wiki/Test-execution-order#test-execution-order, Junit 4 uses deterministic but not predictable order.

Observations:

• Building from scratch doesn't cause any issues. Using distroless and alpine (larger image) causes the next test to hang.
• Building to the docker daemon doesn't cause any issues.
• Additionally, using a local registry at a different port still causes the next test to hang.

[mpeddada1]

[Update on CI] Splitting the test methods up into separate test classes no longer causes the builds to be stuck. Ready for another round of review.

[geoand]

Thanks for the fix folks!

As we are seeing someone reporting something similar in Quarkus, would it be possible to know an approximate timeline for a release of jib-core that will contain the fix? Thanks!

[calebAtIspot]

should https://github.com/GoogleContainerTools/jib/blob/master/docs/faq.md#how-do-i-specify-a-platform-in-the-manifest-list-or-oci-index-of-a-base-image be updated? It still says "Does not support pushing to a Docker daemon"