[dongbo910220]

Purpose

This PR adds an automated test for the behavior fixed in #24897, as requested by @cadedaniel in the original PR's comments.

The new test verifies that the /health endpoint correctly returns 503 Service Unavailable when EngineDeadError is raised, which prevents future regressions of this behavior.

Relates to: #24897

Test Plan

Run the new test case:

pytest tests/entrypoints/openai/test_basic.py::test_health_check_engine_dead_error -v

Test Result

Test passes successfully:

tests/entrypoints/openai/test_basic.py::test_health_check_engine_dead_error PASSED [100%]
======================== 1 passed in 3.55s =========================

The test uses mocks to simulate EngineDeadError being raised during health check, and verifies that the
endpoint correctly returns HTTP 503 status code instead of 500.

Test characteristics:
- Runs quickly (~3-6 seconds) using mocks
- No GPU or model loading required
- Independent of environment-specific configurations
- Focuses specifically on the EngineDeadError → 503 logic

[gemini-code-assist]

Code Review

This pull request introduces a valuable unit test for the /health endpoint. The test correctly verifies that a 503 Service Unavailable status is returned when an EngineDeadError is raised, which is crucial for ensuring the service's health check behaves as expected under failure conditions. The implementation is clean, using mocks effectively to isolate the component under test. This change improves the overall test coverage and helps prevent future regressions. The code is well-structured and I have no concerns.

[dongbo910220]

Hi @DarkLight1337,

Just a friendly follow-up on this PR. Could you please take a look when you have a chance? Any feedback would be greatly appreciated.

Thank you!

[DarkLight1337]

Sorry for the delay, thought @cadedaniel would review it. I'll just stamp this

[dongbo910220]

@DarkLight1337 Great, thank you so much!