开发中常用多线程异步操作场景解决方案

最新推荐文章于 2024-05-17 18:30:12 发布

BACKWASH2038

最新推荐文章于 2024-05-17 18:30:12 发布

阅读量1.2k

点赞数 1

文章标签：后端多线程

本文链接：https://blog.csdn.net/weixin_38501485/article/details/129681547

版权

场景一:批量操作，事务控制部分回滚

第一步:创建现场池，注意不同业务场景线程池队列等参数不一样，根据实际业务场景来

package org.jeecg.modules.pUser.excutor;


import lombok.extern.slf4j.Slf4j;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.scheduling.annotation.AsyncConfigurer;
import org.springframework.scheduling.annotation.EnableAsync;
import org.springframework.scheduling.concurrent.ThreadPoolTaskExecutor;

import java.util.concurrent.ThreadPoolExecutor;

/**
 * @version 1.0
 * @Author zhaozhiqiang
 * @Date 2022/8/18 14:56
 * @Description //TODO  职业病团队批量导入线程池配置
 */
@Slf4j
@Configuration
@EnableAsync// 启用异步任务 151-500
public class PeUserThreadPoolConfig implements AsyncConfigurer {
    public static void main(String[] args) {
        System.out.println(Runtime.getRuntime().availableProcessors());
    }


    /**
     * @version 1.0
     * @Author zhaozhiqiang
     * @Date 2022/8/19 10:59
     * @Description //TODO 团队批量预登记
     */
    @Bean("teamImportExecutor")
    public ThreadPoolTaskExecutor teamImportTaskExecutor() {
        ThreadPoolTaskExecutor threadPoolTaskExecutor = new ThreadPoolTaskExecutor();
        // 设置核心线程数
        threadPoolTaskExecutor.setCorePoolSize(Runtime.getRuntime().availableProcessors());
        // 设置最大线程数
        threadPoolTaskExecutor.setMaxPoolSize(Runtime.getRuntime().availableProcessors() * 2);
        // 设置队列容量,当没有足够的线程去处理任务时，可以将任务放进队列中，以队列先进先出的特性来执行工作任务
        threadPoolTaskExecutor.setQueueCapacity(5000);
        // 设置线程活跃时间（秒）非核心线程数 还回去的时间  如果空闲超过10秒就被回收
        threadPoolTaskExecutor.setKeepAliveSeconds(1);
        // 设置默认线程名称
        threadPoolTaskExecutor.setThreadNamePrefix("teamImport_");
        //用来设置线程池关闭的时候等待所有任务都完成(可以设置时间)
        threadPoolTaskExecutor.setWaitForTasksToCompleteOnShutdown(true);
        //设置上面的时间(秒)
        threadPoolTaskExecutor.setAwaitTerminationSeconds(1);
        //拒绝策略:线程池都忙不过来的时候，可以适当选择拒绝
        /**
         * ThreadPoolExecutor.AbortPolicy();//默认，队列满了丢任务抛出异常
         * ThreadPoolExecutor.DiscardPolicy();//队列满了丢任务不异常
         * ThreadPoolExecutor.DiscardOldestPolicy();//将最早进入队列的任务删，之后再尝试加入队列
         * ThreadPoolExecutor.CallerRunsPolicy();//如果添加到线程池失败，那么主线程会自己去执行该任务
         */
        threadPoolTaskExecutor.setRejectedExecutionHandler(new ThreadPoolExecutor.DiscardPolicy());
        threadPoolTaskExecutor.initialize();
        return threadPoolTaskExecutor;
    }

    /**
     * @version 1.0
     * @Author zhaozhiqiang
     * @Date 2022/8/19 11:00
     * @Description //TODO 批量发布申请单
     */
    @Bean("releaseBatchOrderExecutor")
    public ThreadPoolTaskExecutor releaseBatchOrderTaskExecutor() {
        ThreadPoolTaskExecutor threadPoolTaskExecutor = new ThreadPoolTaskExecutor();
        // 设置核心线程数
        threadPoolTaskExecutor.setCorePoolSize(2);
        // 设置最大线程数
        threadPoolTaskExecutor.setMaxPoolSize(4);
        // 设置队列容量,当没有足够的线程去处理任务时，可以将任务放进队列中，以队列先进先出的特性来执行工作任务
        threadPoolTaskExecutor.setQueueCapacity(30);
        // 设置线程活跃时间（秒）非核心线程数 还回去的时间  如果空闲超过10秒就被回收
        threadPoolTaskExecutor.setKeepAliveSeconds(1);
        // 设置默认线程名称
        threadPoolTaskExecutor.setThreadNamePrefix("releaseBatch_");
        //用来设置线程池关闭的时候等待所有任务都完成(可以设置时间)
        threadPoolTaskExecutor.setWaitForTasksToCompleteOnShutdown(true);
        //设置上面的时间(秒)
        threadPoolTaskExecutor.setAwaitTerminationSeconds(60);
        //拒绝策略:线程池都忙不过来的时候，可以适当选择拒绝
        /**
         * ThreadPoolExecutor.AbortPolicy();//默认，队列满了丢任务抛出异常
         * ThreadPoolExecutor.DiscardPolicy();//队列满了丢任务不异常
         * ThreadPoolExecutor.DiscardOldestPolicy();//将最早进入队列的任务删，之后再尝试加入队列
         * ThreadPoolExecutor.CallerRunsPolicy();//如果添加到线程池失败，那么主线程会自己去执行该任务
         */
        threadPoolTaskExecutor.setRejectedExecutionHandler(new ThreadPoolExecutor.DiscardPolicy());
        threadPoolTaskExecutor.initialize();
        return threadPoolTaskExecutor;
    }
}

第二步:团队批量预登记使用如下(当前场景比较复杂，适用于大批量数据并且事务控制是按照部分回滚的；比如100条中有三条错误那么只回滚这三条的数据)

1 控制层代码
    /**
     * 批量登记
     *
     * @return
     */
    @ApiOperation(value = "批量预登记", notes = "批量预登记")
    @PostMapping(value = "/preBatchRegist")
    public Result<?> preBatchRegist(@RequestParam(value = "ids", required = true) String ids)  {
        String token = UserTokenContext.getToken();
        peTeamImportService.preBatchRegist(ids,token);
        return Result.ok("批量登记成功,结果将陆续执行，请耐心等待所有数据全部执行完成");
    }
2实现层代码
@Override
    @Async("teamImportExecutor")
    public void preBatchRegist(String ids,String token) {
        UserTokenContext.setToken(token);
        if (StringUtils.isBlank(ids)) {
            log.info("批量预登记ids为空");
            return;
        }
        //1 查询数据库组装需要预登记的数据
        List<String> list = Arrays.asList(ids.split(","));
        //将list集合按指定长度进行切分，返回新的List<List<??>>集合
        List<List<String>> idsSplitNumber = Lists.partition(list, BATCH_LIMIT_NUMBER);
        List<PeTeamImportPage> peTeamImportPages = new ArrayList<>();
        idsSplitNumber.forEach(item -> {
            List<PeTeamImport> peTeamImports = this.listByIds(item);
            List<String> selfStrings = new ArrayList<String>();
            peTeamImports.forEach(mainBean -> {
                selfStrings.add(mainBean.getId());
            });
            if (selfStrings.size() > 0) {
                //获取单位导入受检史
                List<PeTeamImportSelf> mainIds = peTeamImportSelfService.getMainIds(selfStrings);
                peTeamImports.forEach(mainItem -> {
                    PeTeamImportPage vo = new PeTeamImportPage();
                    BeanUtils.copyProperties(mainItem, vo);
                    List<PeTeamImportSelf> childDto = new ArrayList<>();
                    mainIds.forEach(childItem -> {
                        if (StringUtils.equals(mainItem.getId(), childItem.getMainId())) {
                            childDto.add(childItem);
                        }
                    });
                    vo.setPeTeamImportSelfList(childDto);
                    peTeamImportPages.add(vo);
                });
            }

        });
        if (peTeamImportPages.size() > 0) {
            //2 开始预登记
            for (int i = 0; i < peTeamImportPages.size(); i++) {
                try {
                    peTeamImportSelfService.prebatchSend(peTeamImportPages.get(i));
                } catch (Exception e) {
                    continue;
                }
            }
        }
    }
3 新事物控制实现类，当前控制新事务，单个有问题单个回滚
@Override
	@Transactional(propagation = Propagation.REQUIRES_NEW,rollbackFor = Exception.class)
	public void prebatchSend(PeTeamImportPage dto) {
		try {
			//预登记之前不能重复登记
			PeTeamImport byId = peTeamImportService.getById(dto.getId());
			if (null != byId && byId.getRegistFlag() == 1) {//已预登记过了不能重复登记
				log.info("编号:{},已经预登记了不能重复登记", dto.getId());
				throw new JeecgBootException("已经预登记了不能重复登记");
			}
			//2 往分组中添加人员信息，必须是为收费的批次，收费是不能添加人员信息的
			LambdaQueryWrapper<PeDepart> departWrapper = new LambdaQueryWrapper<>();
			departWrapper.eq(PeDepart::getDelFlag, 0);
			departWrapper.eq(PeDepart::getId, dto.getDepartId());
			departWrapper.eq(PeDepart::getPayStatus, PeCommonConstant.UNPAID);
			PeDepart departOne = peDepartService.getOne(departWrapper);
			if (null == departOne) {
				log.info("改批次已经缴费或者退费:[{}]", dto.getDepartId());
				throw new JeecgBootException("改批次已经缴费或者退费");
			}
			//3.1分组名称和分组套餐必须先配置好，先判断库里分组是否存在、分组套餐是否已经被启用
			LambdaQueryWrapper<PeDepartGroup> groupWrapper = new LambdaQueryWrapper<>();
			groupWrapper.eq(PeDepartGroup::getDepartId, dto.getDepartId());
			groupWrapper.eq(PeDepartGroup::getGroupName, dto.getGroupName());
			groupWrapper.eq(PeDepartGroup::getDelFlag, 0);
			groupWrapper.eq(PeDepartGroup::getUseFlag, 1);
			groupWrapper.gt(PeDepartGroup::getTotalPrice, 0);//分组价格大于0
			PeDepartGroup groupOne = peDepartGroupService.getOne(groupWrapper);
			if (null == groupOne) {
				log.info(PeCodeConstant.DISABLE_IMPORT_GROUP_MSG);
				throw new JeecgBootException("改分组已经缴费或者退费");
			} else {
				//3.2 分组当前必须配置好套餐有项目可以检查
				LambdaQueryWrapper<PeDepartGroupDetail> groupDetailsWrapper = new LambdaQueryWrapper<>();
				groupDetailsWrapper.eq(PeDepartGroupDetail::getGroupId, groupOne.getId());
				groupDetailsWrapper.eq(PeDepartGroupDetail::getDiseasesFlag, 1);
				groupDetailsWrapper.eq(PeDepartGroupDetail::getDelFlag, 0);
				List<PeDepartGroupDetail> groupDetails = peDepartGroupDetailService.list(groupDetailsWrapper);
				if (null == groupDetails || groupDetails.size() == 0) {//当前分组没有配置套餐无法预登记
					log.info(PeCodeConstant.DISABLE_IMPORT_GROUP_ITEMS_MSG);
					throw new JeecgBootException(PeCodeConstant.DISABLE_IMPORT_GROUP_ITEMS_MSG);
				}
			}
			//4添加登记表、登记详情表、用户表等
			String peNumber = peRegistService.getPeNumber();
			//通过分组编号获取单位批次编号和单位批次名称
			PeUserInfo peUserInfo = null;
			try {
				peUserInfo = new PeUserInfo(0, dto.getName(), dto.getPhone(), null, dto.getDepartId(), dto.getSex(),
						dto.getMarriage(), dto.getIdno(), "" + dto.getAge(), departOne.getCompanyName());
			} catch (Exception e) {
				e.printStackTrace();
				log.info("用户相关信息填写不规范:[{}]", e.getMessage());
				throw new JeecgBootException("用户相关信息填写不规范");
			}
			//保存到体检用户表中
			userInfoService.save(peUserInfo);
			PeRegist build = new PeRegist();
			//工种类型需要转义成字典值
			if (StringUtils.isNotBlank(dto.getHazardType())) {
				String hazardTypeValue = formatHazardType(dto.getHazardType());
				if (StringUtils.isNotBlank(hazardTypeValue)) {
					build.setWorkeType(hazardTypeValue);
				}
			}
			build.setPeType(PeCommonConstant.REAM_TYPE);
			build.setPeNumber(peNumber);
			build.setPeUserId(peUserInfo.getId());
			build.setName(dto.getName());
			build.setOrderTime(departOne.getOrderTime());
			build.setPeDepartId(dto.getDepartId());
			build.setGroupId(groupOne.getId());
			build.setGroupName(dto.getGroupName());
			build.setPackDisFlag(PeCommonConstant.DIS_GROUP_TYPE);
			build.setCurRate(departOne.getDiscount());//单位折扣和个人的保持一直，为了个人和团队统一费用统计核算功能
			build.setSource(PeCommonConstant.WINDOW_REGISTER);
			//转义岗位状态
			if (StringUtils.isNotBlank(dto.getPosStatus())) {
				if (StringUtils.equals(dto.getPosStatus(), PeCommonConstant.POS_STATUS_PRE_STR)) {//上岗前
					build.setPosStatus(PeCommonConstant.POS_STATUS_PRE);
				} else if (StringUtils.equals(dto.getPosStatus(), PeCommonConstant.POS_STATUS_ZG_STR)) {//在岗期间
					build.setPosStatus(PeCommonConstant.POS_STATUS_ZG);
				} else if (StringUtils.equals(dto.getPosStatus(), PeCommonConstant.POS_STATUS_LG_STR)) {//离岗
					build.setPosStatus(PeCommonConstant.POS_STATUS_LG);
				}
			}
			build.setRemarks(PeCommonConstant.DISABLE_IMPORT_REMARKS);
			peRegistService.save(build);
			//成功之后更新导入标识已成功
			PeTeamImport entity = new PeTeamImport();
			entity.setId(dto.getId());
			entity.setRegistFlag(1);
			entity.setRegId(build.getId());
			peTeamImportService.updateById(entity);
			//查询单位批次分组信息
			List<PackgeInfoVo> packgeInfoVos = peDepartGroupDetailService.queryCombInfoById(groupOne.getId());
			addRegistInfo(groupOne.getId(), build, packgeInfoVos);
			//添加当前人员项目对应的所有科室id多个逗号隔开
			String combDepartIds = peRegistMapper.getCombDepartIds(build.getId());
			build.setDepartIds(combDepartIds);
			peRegistService.updateById(build);
			//5添加职业病相关的表
			//插入职业病危害因素种类
			//判断数据库是否已经存在数据
			try {
				//转化空格
				String str = dto.getHazardName().replaceAll("\\s*", "");
				//将中文逗号转化英文逗号
				String finalStr = str.replaceAll("，", ",");
				String finalconsDetailStr = "[" + finalStr + "]";
				PeIndusConsDetailFactor detailFactor = new PeIndusConsDetailFactor(1, dto.getDepartId(), build.getId(), dto.getHazardCategory(), finalconsDetailStr);
				LambdaQueryWrapper<PeIndusConsDetailFactor> wrapper = new LambdaQueryWrapper<>();
				wrapper.eq(PeIndusConsDetailFactor::getDelFlag, 0);
				wrapper.eq(PeIndusConsDetailFactor::getRegId, dto.getId());
				wrapper.eq(PeIndusConsDetailFactor::getIndusDepartId, dto.getDepartId());
				wrapper.eq(PeIndusConsDetailFactor::getDisType, 1);
				wrapper.eq(PeIndusConsDetailFactor::getFactorName, dto.getHazardCategory());
				PeIndusConsDetailFactor detailFactor1 = indusConsDetailFactorMapper.selectOne(wrapper);
				if (null == detailFactor1) {
					if (StringUtils.isNotBlank(dto.getHazardCategory()) && StringUtils.isNotBlank(dto.getHazardName())) {
						indusConsDetailFactorMapper.insert(detailFactor);
					}
				} else {
					detailFactor.setId(detailFactor1.getId());
					indusConsDetailFactorMapper.updateById(detailFactor);
				}
			} catch (Exception e) {
				e.printStackTrace();
				throw new JeecgBootException(e.getMessage());

			}
			//职业历史 indusConsDetailHistoryMapper
			LambdaQueryWrapper<PeIndusConsDetailHistory> historyLambdaQueryWrapper = new LambdaQueryWrapper<>();
			historyLambdaQueryWrapper.eq(PeIndusConsDetailHistory::getRegId, build.getId());
			indusConsDetailHistoryMapper.delete(historyLambdaQueryWrapper);
			//受检本人历史先查看预分组信息中当前人的信息是否存在
			List<PeTeamImportSelf> peTeamImportSelfList = this.selectByMainId(dto.getId());
			if (null != peTeamImportSelfList && peTeamImportSelfList.size() > 0) {
				peTeamImportSelfList.stream().forEach(item -> {
					PeIndusConsDetailHistory history = new PeIndusConsDetailHistory(1, dto.getDepartId(), build.getId(), item.getCycletime(),
							item.getHarmFactor(), item.getProtectMeasures(), item.getUnit(), item.getWorkshop(), item.getWorktype());
					if (StringUtils.isNotBlank(item.getCycletime()) && StringUtils.isNotBlank(item.getHarmFactor()) && StringUtils.isNotBlank(item.getProtectMeasures())
							&& StringUtils.isNotBlank(item.getUnit()) && StringUtils.isNotBlank(item.getWorkshop()) && StringUtils.isNotBlank(item.getWorktype())) {
						indusConsDetailHistoryMapper.insert(history);
					}

				});
			}
			//慢性病历史 indusConsDetailCommonMapper
			LambdaQueryWrapper<PeIndusConsDetailCommon> commonLambdaQueryWrapper = new LambdaQueryWrapper<>();
			commonLambdaQueryWrapper.eq(PeIndusConsDetailCommon::getRegId, build.getId());
			indusConsDetailCommonMapper.delete(commonLambdaQueryWrapper);

			if (StringUtils.isNotBlank(dto.getPastHistory()) || StringUtils.isNotBlank(dto.getSlowName()) || StringUtils.isNotBlank(dto.getOther())) {
				PeIndusConsDetailCommon consDetailCommon = new PeIndusConsDetailCommon(1, dto.getDepartId(), build.getId(), dto.getPastHistory(), dto.getSmokeStatus()
						, dto.getDrinkStatus(), dto.getSmokeAmount(), dto.getSmokeLength(), dto.getFirst(), dto.getCycle(), dto.getPeriod(), dto.getStop(), dto.getChildren(), dto.getFlow(),
						dto.getEarly(), dto.getDie(), dto.getAbnormal(), dto.getOther(), dto.getDrinkAmount(), dto.getDrinkLenth(), dto.getSlowName(), dto.getSlowTime(), dto.getSlowDepart()
						, dto.getHealthFlag());
				indusConsDetailCommonMapper.insert(consDetailCommon);
			}
			//症状 indusConsDetailDiseaseMapper
			LambdaQueryWrapper<PeIndusConsDetailDisease> detailDiseaseLambdaQueryWrapper = new LambdaQueryWrapper<>();
			detailDiseaseLambdaQueryWrapper.eq(PeIndusConsDetailDisease::getIndusDepartId, dto.getDepartId());
			detailDiseaseLambdaQueryWrapper.eq(PeIndusConsDetailDisease::getRegId, build.getId());
			indusConsDetailDiseaseMapper.delete(detailDiseaseLambdaQueryWrapper);
			//转化空格
			String consDetailStr = dto.getSymptom().replaceAll("\\s*", "");
			//将中文逗号转化英文逗号
			String detailStr = consDetailStr.replaceAll("，", ",");
			//用集合包裹起来
			if (StringUtils.isNotBlank(dto.getSymptom())) {
				detailStr = "[" + detailStr + "]";
				PeIndusConsDetailDisease disease = new PeIndusConsDetailDisease(1, dto.getDepartId(), build.getId(), detailStr);
				indusConsDetailDiseaseMapper.insert(disease);
			}
		} catch (Exception e) {
			log.info("异常了:" + e.getMessage());
			throw e;
		}

第三步：批量发布申请单和批量预登记是一个场景，只不过当前的没有上面那个复杂，数据量不是很多

    @Override
    @Async("releaseBatchOrderExecutor")
    public void releaseBatchOrder(String regIds,String token) {
        UserTokenContext.setToken(token);
        if (StringUtils.isNotBlank(regIds)) {
            List<String> list = Arrays.asList(regIds.split(","));
            if (null != list && list.size() > 0) {
                for (int i = 0; i < list.size(); i++) {
                    try {
                        log.info("登记编号：{}", list.get(i));
                        peLisContactService.batchSend(list.get(i));
                    } catch (Exception e) {
                        log.info("异常信息：{}", e.getMessage());
                        continue;
                    }

                }
            }
        }
    }



@Override
    @Transactional(propagation = Propagation.REQUIRES_NEW,rollbackFor = Exception.class)
    public void batchSend(String regId) {
        try {
            //先判断是否lis那边已经推送过了
            String hosId = peChargeMapper.getHosId();
            //业务处理
            log.info("departFeeHandler regId：{} ", regId);
            //同步数据库和lis对接
            List<PeLisContact> peLisContacts = peLisContactMapper.queryListByRegId(regId, hosId);
            log.info("开始执行:{}条", peLisContacts.size());
            if (null != peLisContacts && peLisContacts.size() > 0) {
                log.info("开始插入条数:{}", peLisContacts.size());
                //批量插入
                this.saveBatch(peLisContacts);
            }
        } catch (Exception e) {
            log.info("异常了:" + e.getMessage());
            throw e;
        }
    }

场景二:批量操作，不用控制事务(比如发短信，发飞秋，插入日志等场景)

package org.jeecg.common.util;

import lombok.extern.slf4j.Slf4j;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.scheduling.annotation.AsyncConfigurer;
import org.springframework.scheduling.annotation.EnableAsync;
import org.springframework.scheduling.concurrent.ThreadPoolTaskExecutor;

import java.util.concurrent.ThreadPoolExecutor;

/**
 * @version 1.0
 * @Author zhaozhiqiang
 * @Date 2023/2/13 16:58
 * @Description //TODO 客服线程池
 */
@Slf4j
@Configuration
@EnableAsync// 启用异步任务 151-500
public class CsThreadPoolConfig implements AsyncConfigurer {

    public static void main(String[] args) {
        System.out.println(Runtime.getRuntime().availableProcessors());
    }

    /**
     * @version 1.0
     * @Author zhaozhiqiang
     * @Date 2022/8/19 10:59
     * @Description //TODO频率高将队列容量设置30
     */
    @Bean("feiqiuTask")
    public ThreadPoolTaskExecutor feiqiuTask() {
        ThreadPoolTaskExecutor threadPoolTaskExecutor = new ThreadPoolTaskExecutor();
        // 设置核心线程数
        threadPoolTaskExecutor.setCorePoolSize(Runtime.getRuntime().availableProcessors());
        // 设置最大线程数
        threadPoolTaskExecutor.setMaxPoolSize(Runtime.getRuntime().availableProcessors() * 2);
        // 设置队列容量,当没有足够的线程去处理任务时，可以将任务放进队列中，以队列先进先出的特性来执行工作任务
        threadPoolTaskExecutor.setQueueCapacity(30);
        // 设置线程活跃时间（秒）非核心线程数 还回去的时间  如果空闲超过10秒就被回收
        threadPoolTaskExecutor.setKeepAliveSeconds(1);
        // 设置默认线程名称
        threadPoolTaskExecutor.setThreadNamePrefix("feiqiu_");
        //用来设置线程池关闭的时候等待所有任务都完成(可以设置时间)
        threadPoolTaskExecutor.setWaitForTasksToCompleteOnShutdown(true);
        //设置上面的时间(秒)
        threadPoolTaskExecutor.setAwaitTerminationSeconds(10);
        //拒绝策略:线程池都忙不过来的时候，可以适当选择拒绝
        /**
         * ThreadPoolExecutor.AbortPolicy();//默认，队列满了丢任务抛出异常
         * ThreadPoolExecutor.DiscardPolicy();//队列满了丢任务不异常
         * ThreadPoolExecutor.DiscardOldestPolicy();//将最早进入队列的任务删，之后再尝试加入队列
         * ThreadPoolExecutor.CallerRunsPolicy();//如果添加到线程池失败，那么主线程会自己去执行该任务
         */
        threadPoolTaskExecutor.setRejectedExecutionHandler(new ThreadPoolExecutor.DiscardPolicy());
        threadPoolTaskExecutor.initialize();
        return threadPoolTaskExecutor;
    }

    /**
     * 发送短信
     * 频率高将队列容量设置100
     * @return
     */
    @Bean("smsSendTask")
    public ThreadPoolTaskExecutor smsSendTask() {
        ThreadPoolTaskExecutor threadPoolTaskExecutor = new ThreadPoolTaskExecutor();
        // 设置核心线程数
        threadPoolTaskExecutor.setCorePoolSize(Runtime.getRuntime().availableProcessors());
        // 设置最大线程数
        threadPoolTaskExecutor.setMaxPoolSize(Runtime.getRuntime().availableProcessors() * 2);
        // 设置队列容量,当没有足够的线程去处理任务时，可以将任务放进队列中，以队列先进先出的特性来执行工作任务
        threadPoolTaskExecutor.setQueueCapacity(100);
        // 设置线程活跃时间（秒）非核心线程数 还回去的时间  如果空闲超过10秒就被回收
        threadPoolTaskExecutor.setKeepAliveSeconds(1);
        // 设置默认线程名称
        threadPoolTaskExecutor.setThreadNamePrefix("feiqiu_");
        //用来设置线程池关闭的时候等待所有任务都完成(可以设置时间)
        threadPoolTaskExecutor.setWaitForTasksToCompleteOnShutdown(true);
        //设置上面的时间(秒)
        threadPoolTaskExecutor.setAwaitTerminationSeconds(10);
        //拒绝策略:线程池都忙不过来的时候，可以适当选择拒绝
        /**
         * ThreadPoolExecutor.AbortPolicy();//默认，队列满了丢任务抛出异常
         * ThreadPoolExecutor.DiscardPolicy();//队列满了丢任务不异常
         * ThreadPoolExecutor.DiscardOldestPolicy();//将最早进入队列的任务删，之后再尝试加入队列
         * ThreadPoolExecutor.CallerRunsPolicy();//如果添加到线程池失败，那么主线程会自己去执行该任务
         */
        threadPoolTaskExecutor.setRejectedExecutionHandler(new ThreadPoolExecutor.DiscardPolicy());
        threadPoolTaskExecutor.initialize();
        return threadPoolTaskExecutor;
    }
}

  @Override
    @Async("feiqiuTask")
    public void sendFeiqiuSmg(String title, HttpServletRequest request, String msg, String userId, String token) {
        UserTokenContext.setToken(token);
        //发送feiqiu消息
        String userName = JwtUtil.getUserNameByToken(request);
        LoginUser user = sysUserService.getEncodeUserInfo(userName);
        if (null != user) {
            try {
                SysUser sendUserInfo = sysUserService.getById(userId);
                if (null != sendUserInfo && StringUtils.isNotBlank(sendUserInfo.getClientId())) {
                    FeiQiuUtils.send(title, userName, sendUserInfo.getClientId(), msg);
                    log.info("发送飞秋消息成功:{}", msg);
                }
            } catch (IOException e) {
                e.printStackTrace();
                log.error("发送飞秋消息失败:{}", msg);
            }
        }

    }

场景三:批量操作，控制事务全部回滚(注意:不推荐这样做)

不推荐这样做的原因主要有三

1 要么全部成功要么全部失败，特别耗费连接资源数据量大的时候其它场景操作当前的表的时候等待，有可能操作死表

2 这这场景用的很少，或者几乎不用

3 这种场景后台进程特别耗时，失去了大批量异步的意义

把场景一稍微改造一下就是场景三，循环放到新事物里面就是全部滚回

场景四:批量操作，控制事务遇到错误立马停止，错误之前的不用回滚

当前的场景使用频率还是蛮高的，改造场景一将for循环的continue改成break就完了

BACKWASH2038

关注

1
点赞
踩
3

收藏

觉得还不错? 一键收藏
打赏
0
评论
开发中常用多线程异步操作场景解决方案

第一步:创建现场池，注意不同业务场景线程池队列等参数不一样，根据实际业务场景来第二步:团队批量预登记使用如下(当前场景比较复杂，适用于大批量数据并且事务控制是按照部分回滚的；比如100条中有三条错误那么只回滚这三条的数据)第三步：批量发布申请单和批量预登记是一个场景，只不过当前的没有上面那个复杂，数据量不是很多。
复制链接

扫一扫