[fix] util stop_backend() should always try odemis-stop if nice request fails

[pieleric]

If the backend does stop in time, it would call odemis-stop... however
if the backend didn't even accept the request to stop, it would bail out
immediately.
=> also call odemis-stop in such case. That should help for some
automatic run of test cases

[coderabbitai]

📝 Walkthrough

Walkthrough

The stop_backend function in the testing utility was modified to improve backend termination handling. Instead of raising an exception when the kill command returns a non-zero exit code, the function now treats this as a non-fatal error, marking the backend as DEAD and logging a warning. On successful kill (zero return code), the function implements a polling mechanism that waits up to 15 seconds for the backend to reach a stopped state, checking status repeatedly and logging progress. If the status remains in BUSY state after the timeout, a warning is issued and an optional fallback path attempts force-stop via sudo. Status initialization was added at the start of the success path, and the conditional branches were restructured with updated logging messages.

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly describes the main change: ensuring stop_backend() calls odemis-stop even when the nice request fails.
Description check	✅ Passed	The description explains the problem (backend not accepting stop request) and the solution (also calling odemis-stop in such cases), which aligns with the changeset.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[Copilot]

Pull request overview

Updates the test utility stop_backend() to be more robust by attempting a force-stop (odemis-stop) even when the initial “polite” stop request fails, improving reliability for automated test runs.

Changes:

• Change stop_backend() flow to only wait for shutdown when the initial odemisd --kill succeeds.
• When the odemisd --kill command fails, proceed to the fallback path (instead of raising immediately).

[Copilot]

On non-zero return from the odemisd --kill command, this sets status = driver.BACKEND_DEAD, which forces the later sudo odemis-stop path even when there is actually no backend running (e.g. odemisd --kill exits non-zero with "No running back-end to kill"). Consider setting status = driver.get_backend_status() (or at least checking for BACKEND_STOPPED) so sudo odemis-stop is only attempted when the backend is still running/unresponsive.

[Copilot]

The new behavior for the ret != 0 branch (falling back to sudo odemis-stop when the polite stop request fails) isn’t covered by a test. Adding a unit test that mocks subprocess.call to return non-zero and asserts the correct fallback behavior (and that it avoids calling sudo when get_backend_status() is already BACKEND_STOPPED) would help prevent regressions.