BlockingQueue解决物联网多设备数据整合问题

环境：

　　硬件设备N个采集器将数据以json形式发送至主机，主机将数据以mqtt协议发送至服务端，服务端接收到数据，发布异步事件，进行数据存储，报警等任务。

问题：我们认为不同的传感器在同一个主机上传的数据是同一个设备，数据整合逻辑是，设置一个上传数据间隔

（此间隔认为是，网络延迟+其他延迟），查询语句（select id from xx where deviceid=? and monitortime between ? and ?）,如果有记录，更新，没有记录存储，一次达到合并数据的目的，

此时问题出现，如果在查询时候，语句未走到判断数据存在，那么会出现并发问题，导致数据合并失败。

解决方案：

　　1.并发同步锁

　　2.redis队列

　　3.BlockingQueue

　　4.......

最后选择BlockingQueue解决了这个问题，

主要代码实现：

1.接收到数据，将数据整理成服务端数据格式

2.将BQ给Spring管理

@Configuration
public class QueueAutoConfig {
    @Bean(initMethod = "run")
    public MessageQueueConsume messageQueueConsume(){
        return new MessageQueueConsume(LocalQueue.getMessageQueue());
    }
}

3.BQ实例化

@Slf4j
public class LocalQueue {
    private static BlockingQueue<SaveRecordEvent> messageQueue;
    static {
        messageQueue = new LinkedBlockingQueue<SaveRecordEvent>();
        log.info("queue initialization finished");
    }

    public static BlockingQueue<SaveRecordEvent> getMessageQueue() {
        return messageQueue;
    }
}

4.数据存储逻辑

@Slf4j
public class MessageQueueConsume {
    @Autowired
    IDeviceMonitorRecordService deviceMonitorRecordService;
    @Autowired
    ThreadPoolTaskExecutor asyncServiceExecutor;
    private Long dataRangeTime=90L;
    private BlockingQueue<SaveRecordEvent> blQueue;

    public void run() {
        asyncServiceExecutor.execute(()->{
            while (true) {
                try {
                    SaveRecordEvent saveRecordEvent = blQueue.take();
                    saveRecord(saveRecordEvent);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        });
    }

    private void saveRecord(SaveRecordEvent saveRecordEvent){
        // TODO: 2021/1/7 这里写具体实现 向队列写数据LocalQueue.getMessageQueue().add();
        saveOrUpdate(saveRecordEvent.getEquipmentId(),saveRecordEvent.getDataBean());
    }
    public MessageQueueConsume(BlockingQueue queue) {
        this.blQueue = queue;
    }

    public static void main(String[] args) {
        System.err.println(new Date());
        LocalDateTime now = LocalDateTime.now();
        LocalDateTime localDateTime = now.plusMinutes(1);
        System.err.println(localDateTime);
    }
    public  void saveOrUpdate(String equipmentId, DeviceMonitorRecordDTO bean) {
        LocalDateTime monitorTime = bean.getMonitorTime();
        LambdaQueryWrapper<DeviceMonitorRecordEntity> between = Wrappers.lambdaQuery(DeviceMonitorRecordEntity.class)
                .eq(DeviceMonitorRecordEntity::getEquipmentId, equipmentId)
                .between(DeviceMonitorRecordEntity::getMonitorTime, monitorTime.minusSeconds(dataRangeTime),monitorTime.plusSeconds(dataRangeTime) ).last("limit 1");
        DeviceMonitorRecordEntity one = deviceMonitorRecordService.getOne(between,false);
        // 5.转换db
        DeviceMonitorRecordEntity dbBean = new DeviceMonitorRecordEntity();
        BeanUtil.copyProperties(bean, dbBean);
        // 5.存储逻辑
        if (one == null) {
            // 不存在 保存_
            deviceMonitorRecordService.save(dbBean);
            log.info(monitorTime+"==数据存储成功==");
        } else {
            // 存在更具设别id和时间更新
//            LambdaQueryWrapper<DeviceMonitorRecordEntity> update = Wrappers.lambdaQuery(DeviceMonitorRecordEntity.class)
//                    .eq(DeviceMonitorRecordEntity::getEquipmentId, equipmentId)
//                    .eq(DeviceMonitorRecordEntity::getMonitorTime, one.getMonitorTime());
//            deviceMonitorRecordService.update(dbBean, update);
            dbBean.setId(one.getId());
            deviceMonitorRecordService.updateById(dbBean);
            log.info(monitorTime+"==数据更新成功==");
        }
    }

}

5.发布到BlockingQueue

        //发布阻塞队列之前，将时间更新
        Date date = new Date();
        String time = bean.getTime();
        if (StringUtils.isNotBlank(time)){
            //补数据字段不为空说明此数据是数据补传数据，时间使用此字段
            DateTime parse = DateUtil.parse(time);
            date=parse;
        }
        LocalDateTime localDateTime = DateUtil.toLocalDateTime(date);
        bean.setMonitorTime(localDateTime);
        LocalQueue.getMessageQueue().add(saverecordevent);