超卖问题的4种解决方案、秒杀方案

package com.itsoku.lesson004.service; import com.baomidou.mybatisplus.extension.service.impl.ServiceImpl; import com.itsoku.lesson004.concurrencysafe.ConcurrencyFailException; import com.itsoku.lesson004.concurrencysafe.DbConcurrencySafe; import com.itsoku.lesson004.mapper.GoodsMapper; import com.itsoku.lesson004.po.GoodsPO; import lombok.extern.slf4j.Slf4j; import org.springframework.beans.factory.annotation.Autowired; import org.springframework.stereotype.Service; import org.springframework.transaction.support.TransactionTemplate; import java.util.concurrent.CountDownLatch; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.atomic.AtomicInteger; import java.util.function.Function; /** * <b>description</b>： Java高并发、微服务、性能优化实战案例100讲，，源码 & 文档 & 技术支持， <br> * <b>time</b>：2024/3/26 21:03 <br> * <b>author</b>：ready */ @Service @Slf4j public class GoodsServiceImpl extends ServiceImpl<GoodsMapper, GoodsPO> implements GoodsService { @Autowired private GoodsMapper goodsMapper; @Autowired private DbConcurrencySafe dbConcurrencySafe; @Autowired private TransactionTemplate transactionTemplate; /** * 方案1：通过update中携带条件判断解决超卖问题 * * @throws InterruptedException */ @Override public void placeOrder1() throws InterruptedException { //扣减库存的方法，返回值为1表示扣减库存成功，0表示失败 Function<String, Integer> reduceStock = (String goodsId) -> { int update = goodsMapper.placeOrder1(goodsId, 1); return update; }; //模拟100人秒杀 this.concurrentPlaceOrderMock("方案1", reduceStock); } /** * 方案2：乐观锁解决超卖问题 * * @throws InterruptedException */ @Override public void placeOrder2() throws InterruptedException { //扣减库存的方法，返回值为1表示扣减库存成功，0表示失败 Function<String, Integer> reduceStock = (String goodsId) -> { //1、先查询 GoodsPO goodsPO = this.getById(goodsId); //2、判断库存是否==0，则直接返回失败 if (goodsPO.getNum() == 0) { return 0; } //3.库存看起来够，但是并发的时候可能就不够了，下面带版本号更新库存，判断影响行数，update 为1表示成功，0表示扣减库存失败 int update = goodsMapper.placeOrder2(goodsId, 1, goodsPO.getVersion()); return update; }; //模拟100人秒杀 this.concurrentPlaceOrderMock("方案2", reduceStock); } /** * 方案3：对比数据修改前后是否和期望的一致，解决超卖问题 * * @throws InterruptedException */ @Override public void placeOrder3() throws InterruptedException { //扣减库存的方法，返回值为1表示扣减库存成功，0表示失败 Function<String, Integer> reduceStock = (String goodsId) -> { //1、根据商品id获商品 GoodsPO updateBeforeGoods = this.getById(goodsId); //2、判断库存是否够 if (updateBeforeGoods.getNum() == 0) { return 0; } log.info("库存：{}", updateBeforeGoods.getNum()); //启动事务操作扣减库存 int reduceStockResult = this.transactionTemplate.execute(action -> { //3、执行更新扣减库存 this.goodsMapper.placeOrder3(goodsId, 1); //4、修改数据完成后，查出来看一下，和期望的结果是不是一致的，如果是，表示成功，否则失败 GoodsPO updateAfterGoods = this.getById(goodsId); //5、判断：库存扣减前的数量是否等于 扣减后库存数量+1,如果 if (updateBeforeGoods.getNum() - 1 != updateAfterGoods.getNum()) { //设置事务回滚 action.setRollbackOnly(); return 0; } else { //成功 return 1; } }); return reduceStockResult; }; //模拟100人秒杀 this.concurrentPlaceOrderMock("方案3", reduceStock); } /** * 方案4：通过辅助类解决超卖问题，这种本质上可以解决所有并发修改db数据出错的问题 * * @throws InterruptedException */ @Override public void placeOrder4() throws InterruptedException { //扣减库存的方法，返回值为1表示扣减库存成功，0表示失败 Function<String, Integer> reduceStock = (String goodsId) -> { try { //使用 dbConcurrencySafe.exec 包住需要并发操作的数据，可以确保数据修改的安全性 return this.dbConcurrencySafe.exec(GoodsPO.class, goodsId, () -> { //1、根据商品id获商品 GoodsPO goodsPO = this.getById(goodsId); //2、判断库存是否够 if (goodsPO.getNum() == 0) { return 0; } //3、执行更新扣减库存 this.goodsMapper.placeOrder3(goodsId, 1); return 1; }); } catch (Exception e) { return 0; } }; //模拟100人秒杀 this.concurrentPlaceOrderMock("方案4", reduceStock); } /** * 模拟100人秒杀请求 * * @param fun 扣减库存的函数，fun函数的参数为商品id，返回值：1：表示抢购成功，0：表示抢购失败 * @throws InterruptedException */ private void concurrentPlaceOrderMock(String method, Function<String, Integer> fun) throws InterruptedException { //1、初始化一条商品记录 [商品Id:1,名称：iphone、库存：10] String goodsId = "1", goodsName = "iphone"; // 库存10个 int num = 10; GoodsPO goodsStart = this.initTestData(goodsId, goodsName, num); //2、创建线程池，大小为100，模拟100个线程并发下单 int concurrentNum = 100; ExecutorService executorService = Executors.newFixedThreadPool(concurrentNum); CountDownLatch countDownLatch = new CountDownLatch(concurrentNum); AtomicInteger successNum = new AtomicInteger(0); AtomicInteger failNum = new AtomicInteger(0); //使用线程池模拟100人抢购 for (int i = 0; i < concurrentNum; i++) { executorService.execute(() -> { try { // 调用抢购的函数，update表示抢购的结果，1：表示抢购成功，0：抢购失败 int update = fun.apply(goodsId); //update=0,表示更新失败，否则表示更新成功 if (update == 0) { //抢购失败，失败次数+1，业务上在这里可能要做一些事情（如让事务回滚，那么可以在此抛出一个异常，spring事务会自动回滚） failNum.incrementAndGet(); } else { //更新成功，成功次数+1 successNum.incrementAndGet(); } } finally { countDownLatch.countDown(); } }); } // 调用下面的方法，等待上面线程池中的任务都执行完毕后，才会继续向下走 countDownLatch.await(); // 模拟100人抢购结束，获取抢购完毕后商品的信息 GoodsPO goodsEnd = this.getById(goodsId); //抢购完毕，输出日志，方便