fix(docker): add fail-fast docker socket validation, improve error detection, add tests, and skip integration tests when daemon unavailable

[shivamwayal37]

✨ Description

This PR improves Docker error handling by introducing a fail-fast validation for Unix socket-based Docker hosts.

• Adds pre-validation to detect missing or inaccessible Docker Unix sockets before execution
• Improves detection of Docker socket-related errors by inspecting nested exception causes
• Provides clearer and more actionable error messages when Docker is unavailable
• Adds dedicated unit tests (DockerErrorHandlingTest) to cover these scenarios
• Updates Docker integration tests to skip execution when the Docker daemon/socket is not available

This ensures a clear separation between environment/setup failures and container execution failures, improving reliability and debuggability.

🔗 Related Issue

Closes #15161

🛠️ Backend Checklist

• Code compiles successfully and passes all checks
• All unit and integration tests pass

📝 Additional Notes

Add any extra context or details reviewers should be aware of.

• IllegalStateException is intentionally used for pre-execution environment validation failures instead of TaskException, to avoid mixing distinct failure modes.
• Docker integration tests are conditionally skipped when the Docker daemon is not accessible, preventing false negatives in environments without Docker.

[shivamwayal37]

It looks like the CI failure is unrelated to this PR:

• Failure occurs in BlueprintControllerTest
• Error: Address already in use (port 28181)
• This PR only modifies Docker runner logic

This seems like a flaky test or port conflict in CI.

Would you prefer:

1. I re-run the workflow?
2. Or open a separate PR to make the test use a random port?

Happy to address it separately if needed.

[loicmathieu]

This seems overly complicated.
What I don't unrestand is that we had a code block to handle specifically this kind of exception but it seems it didn't work, maybe fix that code block instead of removing it and replacing it?

[shivamwayal37]

Thanks for the feedback!

You're right — the current approach introduces duplication between the pre-validation and the exception handling block.

My intention was to:

• Fail fast for missing Unix sockets before execution
• Improve detection of Docker socket-related errors from nested exceptions

But I see that this overlaps with the existing logic.

I'll refactor this to:

• Reuse and improve the existing error handling block instead of duplicating it
• Remove redundant checks between pre-validation and catch block

Will push an updated version shortly.

[shivamwayal37]

Thanks for the review @loicmathieu!

I’ve refactored the implementation to remove duplication and simplify the flow:

• Reused and improved the existing error handling instead of replacing it
• Removed redundant checks between pre-validation and the catch block
• Kept fail-fast validation only for early Unix socket detection

This makes the behavior more consistent and easier to follow.

Regarding the CI failure: it occurs in jdbc-postgres (PostgresServiceLivenessCoordinatorTest) and appears unrelated to this PR. The script module (where these changes were made) is passing all tests successfully. This looks like a flaky or environment-related issue, but I’m happy to re-run the workflow or investigate further if needed.

[loicmathieu]

LGTM

[loicmathieu]

Backported to 1.3 and 1.0

[shivamwayal37]

Thanks @loicmathieu and @MilosPaunovic for the review and quick merge!

Glad the refactor aligns well. Appreciate the backport as well 🙌