[yuexia5]

tests/unit/test_core_lifecycle.py::TestAstrBotCoreLifecycleRestart::test_restart_terminates_managers_and_starts_thread hangs forever, so every Unit Tests run now dies on the 30-minute job timeout.

On master all three lanes are cancelled at exactly the timeout:

ubuntu-latest   09:30:54 -> 10:01:08
macos-latest    09:31:40 -> 10:01:58
windows-latest  09:30:46 -> 10:00:59

and the ubuntu log stops at 77%, right before tests/unit/test_core_lifecycle.py:

tests/unit/test_conversation_restart.py ... [ 76%]
............................................................  [ 77%]
<26 minutes of silence, then the job is killed>

Root cause

The test patches astrbot.core.core_lifecycle.threading.Thread:

with patch("astrbot.core.core_lifecycle.threading.Thread") as mock_thread:

astrbot.core.core_lifecycle.threading is the global threading module, so this replaces threading.Thread process-wide.

AstrBotCoreLifecycle.restart() (await added in #10515) then calls:

await asyncio.to_thread(close_trace_store)

asyncio.to_thread() submits to a ThreadPoolExecutor, and its
_adjust_thread_count() spawns workers with threading.Thread. With the mock
in place the "worker" is a MagicMock that never runs, so the callable is never
executed and the awaited future never resolves. The event loop goes idle in
selectors.select and the test hangs forever.

Minimal reproduction, independent of the test suite:

import asyncio
from unittest.mock import patch


def work():
    print("worker ran")


async def main():
    await asyncio.to_thread(work)


with patch("astrbot.core.core_lifecycle.threading.Thread"):
    asyncio.run(asyncio.wait_for(main(), timeout=3))  # never prints "worker ran"

Fix

Patch the threading reference on the lifecycle module instead of
threading.Thread itself, so the real threading module, threading.Thread
and the default executor stay intact. The assertions on the restart thread are
unchanged.

Verification

Before, on master:

$ timeout 60 python -m pytest tests/unit/test_core_lifecycle.py::TestAstrBotCoreLifecycleRestart
<hangs, killed by timeout>

After:

$ python -m pytest tests/unit/test_core_lifecycle.py
26 passed in 2.44s

$ python -m pytest tests
4667 passed, 12 skipped in 163.05s (0:02:43)

The suite previously stalled at 77% for the full 30-minute CI job timeout.

Summary by Sourcery

Fix the lifecycle restart test to isolate its thread mock and prevent executor-backed asynchronous work from hanging the test suite.

Bug Fixes:

• Prevent the lifecycle restart unit test from hanging by avoiding process-wide mocking of Python’s threading implementation.

Tests:

• Update lifecycle restart test mocking so the full unit test suite completes successfully.

[sourcery-ai]

Hey - I've reviewed your changes and they look great!

Sourcery assessment

Approved.

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

[yuexia5]

Closing as a duplicate of #10617, which fixes the same restart-test hang (it mocks asyncio.to_thread in addition to the threading.Thread patch, so the executor worker is never created). #10617 also covers the related trace sn ordering and Windows test issues in the same change set, so keeping both would only split the review. Thanks @wcqqq1214 — I missed it when I opened this.