当前位置：首页 > java >正文

AT模式下的全局锁冲突如何解决？

java 2025/6/23 0:07:06

一、全局锁冲突解决方案

1. 业务层重试机制（推荐方案）

@Service
public class OrderService {@GlobalTransactional@Retryable(maxAttempts = 3, backoff = @Backoff(delay = 100))public void createOrder(OrderDTO order) {// 库存扣减（自动加全局锁）inventoryMapper.reduceStock(order.getProductId(), order.getCount());// 订单创建orderMapper.insert(order);}@Recoverpublic void recoverCreateOrder(OrderException e) {// 告警通知 & 记录异常订单alarmService.send("订单创建失败：" + e.getOrderId());}
}

2. 锁超时配置（框架层优化）

seata:client:lock:retry-interval: 100 # 锁检查间隔(ms)retry-times: 30    # 最大重试次数lock-timeout: 10   # 全局锁等待超时时间(s)

3. 热点数据分离方案

-- 库存表拆分
ALTER TABLE inventory ADD COLUMN sub_stock_1 INT DEFAULT 0;
ALTER TABLE inventory ADD COLUMN sub_stock_2 INT DEFAULT 0;
CREATE INDEX idx_product_sub ON inventory(product_id, sub_index);

public interface InventoryMapper {@Update("UPDATE inventory SET sub_stock_#{subIndex} = sub_stock_#{subIndex} - #{count} " +"WHERE product_id = #{productId} AND sub_stock_#{subIndex} >= #{count}")int reduceSubStock(@Param("productId") String productId, @Param("count") int count,@Param("subIndex") int subIndex);
}

二、高并发优化方案

1. 异步提交事务

@Service
public class AsyncOrderService {@Async("seataAsyncExecutor")@GlobalTransactionalpublic CompletableFuture<Void> asyncCreateOrder(OrderDTO order) {inventoryMapper.reduceStock(order.getProductId(), order.getCount());orderMapper.insert(order);return CompletableFuture.completedFuture(null);}
}

2. 批量操作优化

public interface InventoryMapper {@Update({"<script>","UPDATE inventory SET stock = stock - #{count}","WHERE product_id IN","<foreach item='item' collection='productIds' open='(' separator=',' close=')'>","#{item}","</foreach>","</script>"})int batchReduceStock(@Param("productIds") List<String> productIds, @Param("count") int count);
}