用Worker来保证数据的一致性,再加上MQ来便于水平扩展,也提升了Worker的效率。这就是传说中的Worker+MQ,又叫做可靠消息方式。另外,将任务的查询和执行分工,形成父子任务,达到真正的分布式任务,也能提升Worker效率。
- package com.itlong.bjxizhan.support.web.job.base;
- import com.itlong.bjxizhan.common.DbContext;
- import com.itlong.bjxizhan.domain.pojo.Task;
- import com.itlong.bjxizhan.support.web.service.StandardTaskService;
- import org.slf4j.Logger;
- import org.slf4j.LoggerFactory;
- import java.util.List;
- /**
- * Created by shenhongxi on 2016/7/12.
- */
- public class JobRunnable implements Runnable {
- private static final Logger log = LoggerFactory.getLogger(JobRunnable.class);
- private StandardTaskService standardTaskService;
- private List<Task> tasks;
- private String dbKey;
- private String tableIndex;
- @Override
- public void run() {
- if (tasks != null) {
- try {
- DbContext.setDbKey(dbKey);
- DbContext.setTableIndex(tableIndex);
- for (Task task : tasks) {
- task.setTableIndex(tableIndex);
- // 1. 一个job的多个实例,谁先成功锁定任务,谁先处理任务,若处理失败则解锁任务
- // 2. 对于1中解锁失败的,要利用另外的job来专门进行解锁
- // 3. 将任务分成几批,并行处理
- // 4. 这些任务的子任务分批串行处理,同样有锁定-处理-失败解锁
- // 5. 对于4中解锁失败的,同样要利用另外的job来专门进行解锁
- boolean locked = standardTaskService.lock(task);
- if (!locked) continue;
- boolean result = standardTaskService.process(task);
- standardTaskService.finished(result, task);
- }
- } catch (Exception e) {
- log.error("Do task error", e);
- throw new RuntimeException("Do task error");
- }
- }
- }
- public List<Task> getTasks() {
- return tasks;
- }
- public void setTasks(List<Task> tasks) {
- this.tasks = tasks;
- }
- public StandardTaskService getStandardTaskService() {
- return standardTaskService;
- }
- public void setStandardTaskService(StandardTaskService standardTaskService) {
- this.standardTaskService = standardTaskService;
- }
- public String getTableIndex() {
- return tableIndex;
- }
- public void setTableIndex(String tableIndex) {
- this.tableIndex = tableIndex;
- }
- public String getDbKey() {
- return dbKey;
- }
- public void setDbKey(String dbKey) {
- this.dbKey = dbKey;
- }
- }
***:org.quartz.Scheduler提供了start, bystand等方法可以对Scheduler进行管控