缓存问题
核心是如何避免大量请求到达数据库
缓存穿透
既不存在于 redis，也不存在于 mysql 的key，被重复请求
public Result queryById(Long id) {String key = CACHE_SHOP_KEY+id;// 1. redis & mysqlString shopJson = stringRedisTemplate.opsForValue().get(key);if(StrUtil.isNotBlank(shopJson)){ // 非空字符串, 如"abc"Shop shop = JSONUtil.toBean(shopJson, Shop.class);return Result.ok(shop);}// 空字符串 ""if (shopJson != null){return Result.fail("not found shop");}// 2. !redis & !mysqlShop shop = shopMapper.selectById(id);if(shop == null){// 缓存 value=null 的对象stringRedisTemplate.opsForValue().set(key, "", CACHE_NULL_TTL, TimeUnit.MINUTES);return Result.fail("not fount shop");}// 3. !redis & mysqlstringRedisTemplate.opsForValue().set(key, JSONUtil.toJsonStr(shop), CACHE_SHOP_TTL, TimeUnit.MINUTES);return Result.ok(shop);
}
缓存雪崩
大量key失效（可能是刚好过期，也可能是 Redis 宕机了）
- 给不同的 key 添加 TTL 随机值
- 使用redis集群
- springcloud 分级缓存
缓存击穿
高并发的单key，某种原因过期了
Solution：互斥锁，逻辑过期
视频里用的是递归，我用了 do while，目前是没有问题的
public Result queryById(Long id) {String key = CACHE_SHOP_KEY+id;String lockKey = LOCK_SHOP_KEY+id;try {do {// 查 redisString shopJson = stringRedisTemplate.opsForValue().get(key);if(StrUtil.isNotBlank(shopJson)){ // 非空字符串, 如"abc"Shop shop = JSONUtil.toBean(shopJson, Shop.class);return Result.ok(shop);}if (shopJson != null){return Result.fail("not found shop");}boolean lock = tryLock(lockKey);if(!lock){Thread.sleep(50);} else{break;} }while (true);// 2. !redis & !mysqlShop shop = shopMapper.selectById(id);if(shop == null){// 缓存 value=null 的对象stringRedisTemplate.opsForValue().set(key, "", CACHE_NULL_TTL, TimeUnit.MINUTES);return Result.fail("not fount shop");}// 3. !redis & mysqlstringRedisTemplate.opsForValue().set(key, JSONUtil.toJsonStr(shop), CACHE_SHOP_TTL, TimeUnit.MINUTES);return Result.ok(shop);}catch (InterruptedException e) {throw new RuntimeException(e);}finally {unlock(lockKey);}
}
线程安全
超卖，一人一单存在线程问题
@Transactional
public Result seckillVoucher(long voucherId) {// 查询优惠券SeckillVoucher voucher = secKillVoucherService.getById(voucherId);// 秒杀时间if(voucher.getBeginTime().isAfter(LocalDateTime.now())){return Result.fail("秒杀尚未开始");}if(voucher.getEndTime().isBefore(LocalDateTime.now())){return Result.fail("秒杀已经结束");}// 库存是否足够if(voucher.getStock()<1)return Result.fail("库存不足");// 一人一单检查Long userId = UserHolder.getUser().getId();Long count = query().eq("user_id", userId).eq("voucher_id", voucherId).count();if(count > 0){return Result.fail("用户已经购买过了");}// 减库存, update table set stock=stock-1 where id=1 and stock>0boolean flag = secKillVoucherService.update().setSql("stock=stock-1").eq("voucher_id", voucherId).gt("stock", 0).update();if (flag == false) return Result.fail("库存不足");// 创建订单VoucherOrder voucherOrder = new VoucherOrder();long orderId = redisIdWorker.nextId("order");voucherOrder.setId(orderId); // order idvoucherOrder.setUser_id(userId); // user idvoucherOrder.setVoucher_id(voucherId);save(voucherOrder);return Result.ok(orderId);
}
单机环境下的悲观锁，实现一人一单
public Result seckillVoucher(long voucherId) {//查询优惠券SeckillVoucher voucher = secKillVoucherService.getById(voucherId);// 秒杀时间if(voucher.getBeginTime().isAfter(LocalDateTime.now())){return Result.fail("秒杀尚未开始");}if(voucher.getEndTime().isBefore(LocalDateTime.now())){return Result.fail("秒杀已经结束");}// 库存是否足够if(voucher.getStock()<1)return Result.fail("库存不足");Long userId = UserHolder.getUser().getId();// 获取锁, 函数内部:提交事务(减库存,建订单), 释放锁synchronized (userId.toString().intern()){ // 按值加锁: 锁相同用户的多次请求, 而不是锁方法// 获取代理对象(事务)VoucherOrderService proxy = (VoucherOrderService) AopContext.currentProxy();return proxy.createSeckillOrder(voucherId);}}
@Transactional
public Result createSeckillOrder(long voucherId) {Long userId = UserHolder.getUser().getId();Long count = query().eq("user_id", userId).eq("voucher_id", voucherId).count();if(count > 0){return Result.fail("用户已经购买过了");}// 减库存, update table set stock=stock-1 where id=1 and stock>0boolean flag = secKillVoucherService.update().setSql("stock=stock-1").eq("voucher_id", voucherId).gt("stock", 0).update();if (flag == false) return Result.fail("库存不足");// 创建订单VoucherOrder voucherOrder = new VoucherOrder();long orderId = redisIdWorker.nextId("order");voucherOrder.setId(orderId); // order idvoucherOrder.setUser_id(userId); // user idvoucherOrder.setVoucher_id(voucherId);save(voucherOrder);return Result.ok(orderId);
}
分布式锁
通过 redis 给不同主机加锁，给锁加标识，避免被其他主机释放
问题1 锁误删情况如下：
改进后如下：
问题二：判断锁是自己创建的，到删除锁这段时间发生了阻塞，导致删除别人的锁，因为key都是一样的，判断是根据value
使用 redis lua 脚本，处理判断和删除的逻辑
