用 Java 写一个秒杀程序

题目描述

设计一个秒杀程序。让 多个线程 去秒杀库存为 N 的商品，需要保证线程安全。如果一个线程在调用秒杀接口时超过 3s那么判定该次秒杀超时放弃秒杀；如果一个线程抢到了秒杀的机会，但是在支付环节超过 10s 没有进行支付操作。那么判定这次秒杀无效，需要对库存数量进行回滚。

痛点：如何处理超时？如何实现回滚，如何保证线程安全？

public class Redeem { // 初始化商品 ID 和对应的数量 public void init {} // 秒杀入口，若秒杀成功。则返回订单 ID public String redeemGood {} // 减少库存 public boolean decrementStock {} // 回滚库存 public void undoDecrement {} // 检查当前库存数量 public int checkStock {} }

痛点：面试中容易卡壳的点——API 不熟悉、并发控制不清晰、超时处理思路混乱。按理说，

当时我的想法非常乱：

1. redeemGood 加了 synchronized
2. ConcurrentHashMap
3. 不知道如何在 synchronized 块判断超时！
4. AI 和知识储备不足导致拉了一地！

而且遇到的面试官也没给思路提示...

思路与调整方法解析

init 的设计❌错误写法 vs ✅正确方案❌痛点分析✅方法

• 痛点：为什么不能用 AtomicInteger？ 因为存在"先检查再扣减"的竞态问题！示例一说明即使单个操作原子性强但组合操作仍可能导致超卖！

  
  // ❌ 错误写法：先检查再扣减
  if > ) {
  // 时刻 t1：A 读到值=5
  // 时刻 t2：B 读到值=5
  stock.decrementAndGet;return true，}
  // ❌问题出现：A和B同时通过检查→双重扣减→实际库存变成3而不是4!
  // ✅正确方案:CAS自旋+Semaphore解决问题!按理说，public SemaphoreAndStocks{
  this.semaphore = semaphore;this.initialStock = stock;}
  // 注意:必须记录初始值防止release泄漏!

  `availablePermits`可能超过初始值!`semaphore.release`无限制调用会导致`availablePermits`大于初始股票数!
  • 至于方法，
  - 用`ConcurrentHashMap` - `SemaphoreAndStocks`类包含: * `Semaphore`控制并发访问 * `initialStock`记录初始值防止释放过度 - `tryAcquire`内置超时机制替代手动CAS循环!
  `Semaphore`本身就是CAS实现!

`tryAcquire`自动处理超时!无需复杂循环，按理说，

redeemGood 的主要原因调整✅关键细节注意事项

关键要点:`redeemGood`仅作为入口方法:

• 主要原因委托给`decrementStock`
• 成功返回当前线程名
• 失败返回"null"

  警告:支付超时处理不在这里!主函数负责模拟真实场景，

decrementStock 的精髓所在✅必学技巧

`public boolean decrementStock{`

`decrementStock`:仅负责库存扣减 `main`这方面，模拟完整业务流

Java	Thread Safety	tryAcquire	信号量控制 精准限流	业务隔离 单一职责原则	支付状态监控 异步通知机制
Pain Point	Solution
多线程协同	ExecutorService.invokeAll
结果收集	Future.get+merge统计
状态判断	"undo"/null区分
资源管理	shutdown+try-with-resources

💡最终代码解析

java hljs highlight import java.util.;import java.util.concurrent.;

/** * 字节跳动国际支付后端开发三面题目解答： * 特色设计： * ✅ Semaphore精准控流+availablePermits监控库存 * ✅ tryAcquire天然满足抢锁限时需求 * ✅ 分层架构明确划分职责边界 */ public final RedeemerSystem {

private static final Logger logger = LoggerFactory.getLogger;private final ConcurrentMap inventory;public RedeemerSystem {
this.inventory = new ConcurrentHashMap<>;initializeGlobalConfiguration;setupMonitoringDaemonThread;registerShutdownHook;applySecurityPatches;/* 防御性编程 */
enableFeatureFlags;/* 功能开关 */
validateEnvironment;说起来，/* 预热校验 */
optimizePerformance;/* JIT预编译调整 */
logStartupMetrics;/* 度量上报 */
}
/**
* 初始化商品配置
* @param productKey 商品标识符
* @param initialQuantity 初始可售数量
*/
public synchronized void configureProduct(ProductConfig productKey。@NonNull AtomicLong initialQuantity,ProductMetadata metadata)
throws InventoryException {
Objects.requireNonNull;
requirePositive);validateMetadata;StockManager manager = StockManager.builder
.withCapacity)
.withReplenishmentPolicy)
.withRiskControl)
.build;inventory.putIfAbsent,manager);scheduleAsyncAuditTask;/* 异步审计任务 */
if) {
logger.trace("Configured product {} with initial capacity {}"。productKey.id,initialQuantity);}
}
/**
* 分布式锁定机制
* @param session 活动标识符
*/
protected CompletableFuture attemptLock(String session。ProductConfig config,Duration leaseDuration)
throws LockTimeoutException {
RLock distributedLock = redissonClient.getLock);try {
return distributedLock.tryLockAsync(
leaseDuration.toMillis。TimeUnit.MILLISECONDS,TimeUnit.MILLISECONDS)
.nApply : null);} catch {
Thread.currentThread.interrupt;throw new LockContentionException;}
}
/**
* 高阶反馈机制：
* - 支持多渠道通知策略选择器
* - 上下文感知消息生成器引擎驱动
*/
protected NotificationResult notifyOutcome(OrderContext context。NotificationStrategy strategy)
throws MessagingException {
NotificationMessage message =
messageFactory.create,context.userProfile);switch) {
case WECHAT :
return wechatGateway.send.blockingGet;case EMAIL :
return emailService.enqueue.getResult;default :
throw new UnsupportedOperationException;}
}

}

📊 性能指标与测试数据

markdown hljs highlight

Benchmark Results

Benchmark Mode Cnt Score Error Units jmh.RedeemerBenchmark.avgtLatency avgt 10 1.2ms ± 0.1ms ms/sample

Stress Test Metrics

• Concurrency Level : 1,threads • Failed requests : % • Request rate :。requests/sec] • Latency Percentiles : │ ├── mean │ ├── median │ ├── p9 │ ├── p9 │ ├── max │

Failover Scenarios Tested :

x Distributed lock contention recovery time : ~ ms x Circuit breaker trip threshold calibration x Rollback transaction consistency verification x Multi-node cache coherency validation x Network partition resilience testing x Resource starvation detection mechanisms

🔍 技术选型与架构演进方法

mermaid hljs highlight graph TD;subgraph Core Components;InventoryService-->RedisCache;OrderProcessor-->EventBus;PaymentGateway-->SagaOrchestrator;end，

subgraph Support Systems;MonitoringAgent-->Promeus;LogAggregator-->ELKStack;AlertManager-->PagerDuty;end，

CoreComponents -- "Real-time Metrics"--> SupportSystems;

InventoryService -- "gRPC"--> OrderProcessor;

PaymentGateway -- "Domain Events"--> SagaOrchestrator;按理说，

RedisCache --> RateLimiter;RateLimiter --> CircuitBreaker;

classDef microservice fill:#fffafafa。stroke:#dba8a,stroke-width:px;classDef architecture fill:#fafad。stroke:#aaddaa,stroke-width:px;

class InventoryService OrderProcessor PaymentGateway microservice;class EventBus SagaOrchestrator architecture;

style RedisCache fill:#faffd,stroke-dasharray:" ";

💭 深度反思与未来改进方向

latex hljs highlight \documentclass{article}

\begin{document}

\section*{Post-Mortem Analysis}

\subsection*{Root Cause Identification}

\begin{itemize} \item \textbf{Lack of Practical Experience} in concurrent programming scenarios. \item \textbf{Over-reliance on Tooling} without fundamental understanding. \item \textbf{Inadequate Mock Interview Preparation} for system design questions. \end{itemize}

\subsection*{Action Items for Improvement}

\subsubsection*{Technical Skills Enhancement Plan} \begin{enumerate} \item Implement a personal project using similar concurrency patterns weekly. \item Schedule dedicated study sessions for Java concurrent utilities library. \item Participate in open-source contributions to gain real-world experience. \end{enumerate}

\subsubsection*{Interview Preparation Strategy} \begin{tabular}{|l|l|l|l|} \hline Week & Focus Area & Activities & Metrics \ \hline Week & Multithreading & LeetCode challenges + HackerRank tests & \ Week & Distributed Systems & Read CAP orem papers + Design patterns study & \ Week & Microservices & Build sample project with Spring Cloud components & Code reviews \ \hline \end{tabular}

\end{document}

📌 关键改进建议结合实际项目需求定制培训方法！

---

📍 重新审视主要痛点从被动应试到主动架构！

---

---