feat(adapters): 实现 CircuitBreaker 内存级熔断器
注入 now 纯确定性,force_open 支持 401/403 直接熔断。 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -0,0 +1,70 @@
|
|||||||
|
"""进程内熔断器:按 source_name 记连续失败数与冷却截止时刻。
|
||||||
|
|
||||||
|
内存级单实例,now 由调用方注入以保证纯确定性可测试。达阈值后开路冷却,
|
||||||
|
冷却到期放行试探(half-open)。force_open 支持 401/403 一次即熔断。
|
||||||
|
"""
|
||||||
|
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
|
||||||
|
class CircuitBreaker:
|
||||||
|
"""按 source_name 记失败数;达阈值后开路冷却,冷却到期放行试探(half-open)。
|
||||||
|
|
||||||
|
Args:
|
||||||
|
fail_threshold: 连续失败次数阈值,达到后触发开路。
|
||||||
|
cooldown_s: 开路冷却时长(秒),冷却到期自动进入半开状态。
|
||||||
|
"""
|
||||||
|
|
||||||
|
def __init__(self, fail_threshold: int, cooldown_s: float) -> None:
|
||||||
|
self._fail_threshold = fail_threshold
|
||||||
|
self._cooldown_s = cooldown_s
|
||||||
|
self._fails: dict[str, int] = {}
|
||||||
|
self._open_until: dict[str, float] = {}
|
||||||
|
|
||||||
|
def is_open(self, source_name: str, now: float) -> bool:
|
||||||
|
"""判断指定源是否处于开路状态。
|
||||||
|
|
||||||
|
冷却截止时刻之前为开路;到期返回 False(放行一个试探,即半开)。
|
||||||
|
|
||||||
|
Args:
|
||||||
|
source_name: 被熔断的源标识。
|
||||||
|
now: 当前时刻(秒级时间戳),由调用方注入。
|
||||||
|
|
||||||
|
Returns:
|
||||||
|
True 表示开路(拒绝请求),False 表示关闭或半开(放行)。
|
||||||
|
"""
|
||||||
|
until = self._open_until.get(source_name)
|
||||||
|
return until is not None and now < until
|
||||||
|
|
||||||
|
def record_failure(self, source_name: str, now: float) -> None:
|
||||||
|
"""记录一次失败;累计达阈值则开路至 now + cooldown。
|
||||||
|
|
||||||
|
Args:
|
||||||
|
source_name: 失败的源标识。
|
||||||
|
now: 当前时刻(秒级时间戳)。
|
||||||
|
"""
|
||||||
|
count = self._fails.get(source_name, 0) + 1
|
||||||
|
self._fails[source_name] = count
|
||||||
|
if count >= self._fail_threshold:
|
||||||
|
self._open_until[source_name] = now + self._cooldown_s
|
||||||
|
|
||||||
|
def force_open(self, source_name: str, now: float) -> None:
|
||||||
|
"""强制开路(用于 401/403 等不可恢复错误),一次即熔断。
|
||||||
|
|
||||||
|
将失败计数预设为阈值,使半开探针仅需一次失败即可重新熔断。
|
||||||
|
|
||||||
|
Args:
|
||||||
|
source_name: 需要强制熔断的源标识。
|
||||||
|
now: 当前时刻(秒级时间戳)。
|
||||||
|
"""
|
||||||
|
self._fails[source_name] = self._fail_threshold
|
||||||
|
self._open_until[source_name] = now + self._cooldown_s
|
||||||
|
|
||||||
|
def record_success(self, source_name: str) -> None:
|
||||||
|
"""记录一次成功;清零失败计数与开路状态(关闭熔断器)。
|
||||||
|
|
||||||
|
Args:
|
||||||
|
source_name: 成功的源标识。
|
||||||
|
"""
|
||||||
|
self._fails.pop(source_name, None)
|
||||||
|
self._open_until.pop(source_name, None)
|
||||||
@@ -0,0 +1,69 @@
|
|||||||
|
"""CircuitBreaker 单元测试。
|
||||||
|
|
||||||
|
验证内存级熔断器的核心行为:阈值触发、冷却半开、成功重置、强制熔断。
|
||||||
|
"""
|
||||||
|
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
import pytest
|
||||||
|
|
||||||
|
from adapters.breaker import CircuitBreaker
|
||||||
|
|
||||||
|
|
||||||
|
@pytest.fixture()
|
||||||
|
def breaker() -> CircuitBreaker:
|
||||||
|
"""创建 threshold=3, cooldown=10s 的熔断器。"""
|
||||||
|
return CircuitBreaker(fail_threshold=3, cooldown_s=10.0)
|
||||||
|
|
||||||
|
|
||||||
|
class TestCircuitBreaker:
|
||||||
|
"""CircuitBreaker 行为验证。"""
|
||||||
|
|
||||||
|
def test_closed_by_default(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""初始状态为关闭(放行)。"""
|
||||||
|
assert breaker.is_open("llm", now=0.0) is False
|
||||||
|
|
||||||
|
def test_opens_after_threshold_failures(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""连续失败未达阈值不开路;达到阈值后开路。"""
|
||||||
|
# 2 次失败 < threshold=3,仍关闭
|
||||||
|
breaker.record_failure("llm", now=1.0)
|
||||||
|
breaker.record_failure("llm", now=2.0)
|
||||||
|
assert breaker.is_open("llm", now=2.5) is False
|
||||||
|
|
||||||
|
# 第 3 次失败 >= threshold=3,开路
|
||||||
|
breaker.record_failure("llm", now=3.0)
|
||||||
|
assert breaker.is_open("llm", now=3.5) is True
|
||||||
|
|
||||||
|
def test_half_open_after_cooldown(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""冷却期过后自动半开(放行试探)。"""
|
||||||
|
for t in range(3):
|
||||||
|
breaker.record_failure("llm", now=float(t))
|
||||||
|
# 开路期间
|
||||||
|
assert breaker.is_open("llm", now=5.0) is True
|
||||||
|
# cooldown=10s,第 3 次失败 now=2.0,open_until=12.0
|
||||||
|
# now=12.0 时冷却到期 → 半开
|
||||||
|
assert breaker.is_open("llm", now=12.0) is False
|
||||||
|
|
||||||
|
def test_success_resets(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""record_success 重置计数并关闭熔断器。"""
|
||||||
|
for t in range(3):
|
||||||
|
breaker.record_failure("llm", now=float(t))
|
||||||
|
assert breaker.is_open("llm", now=3.0) is True
|
||||||
|
|
||||||
|
breaker.record_success("llm")
|
||||||
|
assert breaker.is_open("llm", now=3.0) is False
|
||||||
|
|
||||||
|
def test_force_open_immediate(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""force_open 无需积累失败即可直接熔断。"""
|
||||||
|
breaker.force_open("llm", now=5.0)
|
||||||
|
assert breaker.is_open("llm", now=5.5) is True
|
||||||
|
|
||||||
|
def test_force_open_probe_failure_reopens(self, breaker: CircuitBreaker) -> None:
|
||||||
|
"""半开探针失败后重新熔断(因 fails 已置为 threshold)。"""
|
||||||
|
breaker.force_open("llm", now=0.0)
|
||||||
|
# cooldown=10s → open_until=10.0,now=10.0 半开
|
||||||
|
assert breaker.is_open("llm", now=10.0) is False
|
||||||
|
|
||||||
|
# 半开探针失败:一次 record_failure 即达阈值(fails 已=3),重新熔断
|
||||||
|
breaker.record_failure("llm", now=10.0)
|
||||||
|
assert breaker.is_open("llm", now=10.5) is True
|
||||||
Reference in New Issue
Block a user