中断处理机制解析：Claude在异常流程中的最佳实践

2次阅读

共计 2578 个字符，预计需要花费 7 分钟才能阅读完成。

在分布式系统设计中，CAP 理论指出网络分区 (Partition tolerance) 发生时，必须在一致性 (Consistency) 和可用性 (Availability) 间作出抉择。服务中断的典型表现包括：

重复请求(Duplicate Requests)：客户端超时重试导致服务端重复处理
部分成功(Partial Success)：多节点间状态不一致，如支付成功但订单未更新
状态丢失(State Loss)：中间处理结果因进程崩溃未能持久化

处理方案	优点	缺点	适用场景
传统重试	实现简单	无法解决幂等性问题	短时临时故障
事务日志	可追溯完整操作链	日志膨胀影响性能	金融类强一致性系统
事件溯源	支持任意时间点状态重建	学习曲线陡峭	审计严格的业务系统

import time
import threading

class SnowflakeGenerator:
    def __init__(self, datacenter_id: int, worker_id: int):
        self.epoch = 1609459200000  # 2021-01-01 UTC
        self.datacenter_id = datacenter_id
        self.worker_id = worker_id
        self.sequence = 0
        self.last_timestamp = -1
        self.lock = threading.Lock()

    def generate_id(self) -> int:
        with self.lock:
            timestamp = self._current_millis()

            if timestamp < self.last_timestamp:
                raise ValueError("Clock moved backwards")

            if timestamp == self.last_timestamp:
                self.sequence = (self.sequence + 1) & 0xFFF
                if self.sequence == 0:
                    timestamp = self._wait_next_millis(self.last_timestamp)
            else:
                self.sequence = 0

            self.last_timestamp = timestamp

            return ((timestamp - self.epoch) << 22) | \
                   (self.datacenter_id << 17) | \
                   (self.worker_id << 12) | \
                   self.sequence

    def _current_millis(self) -> int:
        return int(time.time() * 1000)

    def _wait_next_millis(self, last_timestamp: int) -> int:
        timestamp = self._current_millis()
        while timestamp <= last_timestamp:
            timestamp = self._current_millis()
        return timestamp

type CompensationTx struct {
    TxID        string
    MainAction  func() error
    Compensate  func() error
    Status      string // "init", "executing", "done", "failed"
}

func (c *CompensationTx) Execute() error {
    c.Status = "executing"
    if err := c.MainAction(); err != nil {if compensateErr := c.Compensate(); compensateErr != nil {
            return fmt.Errorf("main error: %v, compensate error: %v", 
                err, compensateErr)
        }
        c.Status = "failed"
        return err
    }
    c.Status = "done"
    return nil
}

@startuml
participant Client
participant "API Gateway" as Gateway
participant "Idempotency Service" as Idempotency
participant "State Manager" as State
participant "Event Store" as Events

Client -> Gateway: 请求(含 requestId)
Gateway -> Idempotency: 检查 requestId
alt 已处理过
    Idempotency --> Gateway: 返回缓存结果
else 新请求
    Gateway -> State: 锁定资源
    State --> Gateway: 资源令牌
    Gateway -> Events: 持久化事件
    Events --> Gateway: 确认持久化
    Gateway -> State: 更新状态
    State --> Gateway: 新状态版本
    Gateway -> Idempotency: 缓存结果
end
Gateway --> Client: 响应
@enduml