quartz的一些记录

定时任务总会遇到任务重叠执行的情况，比如一个任务1分钟执行一次，而任务的执行时间超过了1分钟，这样就会有两个相同任务并发执行了。有时候我们是允许这种情况的发生的，比如任务执行的代码是幂等的，而有时候我们可能考虑到一些情况是不允许这种事情发生的。

在实际场景中，我们定时任务调度使用quartz来实现触发的，定时任务的业务代码分布在各个应用，用soa调用。

对于quartz来说，官方文档上明确对这种需求有指定的解决办法，就是使用注解@DisallowConcurrentExecution；

意思是：禁止并发执行多个相同定义的JobDetail，就是我们想要的。

下面一个实现的例子：可以对比两个job：AllowConcurrentExecutionTestJob，DisallowConcurrentExecutionTestJob

public class AllowConcurrentExecutionTestJob implements Job {
    public AllowConcurrentExecutionTestJob() {
    }
 
    public void execute(JobExecutionContext context) throws JobExecutionException {
 
        try {
            List<JobExecutionContext> list = context.getScheduler().getCurrentlyExecutingJobs();
            for(JobExecutionContext jobExecutionContext : list){
                // job内部获取容器内变量
                System.out.println(jobExecutionContext.getJobDetail().getKey().getName());
            }
            Thread.sleep(4000);
        } catch (SchedulerException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        System.out.println("Hello World!  AllowConcurrentExecutionTestJob is executing.");
    }
}

@DisallowConcurrentExecution
public class DisallowConcurrentExecutionTestJob implements org.quartz.Job {
    public DisallowConcurrentExecutionTestJob() {
    }
 
    public void execute(JobExecutionContext context) throws JobExecutionException {
 
        try {
            List<JobExecutionContext> list = context.getScheduler().getCurrentlyExecutingJobs();
            for(JobExecutionContext jobExecutionContext : list){
                // job内部获取容器内变量
                System.out.println(jobExecutionContext.getJobDetail().getKey().getName());
            }
            Thread.sleep(4000);
        } catch (SchedulerException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        System.out.println("Hello World!  DisallowConcurrentExecutionTestJob is executing.");
    }
}

测试代码：

public class QuartzTest {
    public static void main(String[] args) throws InterruptedException {
 
        try {
            // Grab the Scheduler instance from the Factory
            Scheduler scheduler = StdSchedulerFactory.getDefaultScheduler();
 
            // and start it off
            scheduler.start();
 
            // define the job and tie it to our HelloJob class
            JobDetail job = JobBuilder.newJob(DisallowConcurrentExecutionTestJob.class)
                    .withIdentity("job1", "group1")
                    .build();
 
            // Trigger the job to run now, and then repeat every 40 seconds
            Trigger trigger = TriggerBuilder.newTrigger()
                    .withIdentity("trigger1", "group1")
                    .startNow()
                    .withSchedule(SimpleScheduleBuilder.simpleSchedule()
                            .withIntervalInSeconds(1)
                            .repeatForever())
                    .build();
 
            // define the job and tie it to our HelloJob class
            JobDetail job2 = JobBuilder.newJob(AllowConcurrentExecutionTestJob.class)
                    .withIdentity("job2", "group1")
                    .build();
 
            // Trigger the job to run now, and then repeat every 40 seconds
            Trigger trigger2 = TriggerBuilder.newTrigger()
                    .withIdentity("trigger2", "group1")
                    .startNow()
                    .withSchedule(SimpleScheduleBuilder.simpleSchedule()
                            .withIntervalInSeconds(1)
                            .repeatForever())
                    .build();
 
            // Tell quartz to schedule the job using our trigger
            scheduler.scheduleJob(job2, trigger2);
//            scheduler.scheduleJob(job2, trigger2);
            // wait trigger
            Thread.sleep(20000);
            scheduler.shutdown();
 
        } catch (SchedulerException se) {
            se.printStackTrace();
        }
    }
}

我们还发现在job的execute里传参是JobExecutionContext，它可以让我们拿到正在执行的job的信息。所以我想在job里直接判断一下就可以知道有没有已经在执行的相同job。

public class SelfDisAllowConExeTestJob implements org.quartz.Job{
    public void execute(JobExecutionContext context) throws JobExecutionException {
        try {
            List<JobExecutionContext> list = context.getScheduler().getCurrentlyExecutingJobs();
            Set<String> jobs = new HashSet<String>();
            int i=0;
            for (JobExecutionContext jobExecutionContext : list){
                if(context.getJobDetail().getKey().getName().equals(jobExecutionContext.getJobDetail().getKey().getName())){
                    i++;
                }
            }
            if(i>1){
                System.out.printf("self disallow ");
                return;
            }
            Thread.sleep(4000);
        } catch (SchedulerException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        System.out.println("Hello World!  SelfDisAllowConExeTestJob is executing.");
 
    }
}

测试代码：

public class OuartzSelfMapTest {
 
    public static void main(String[] args) throws InterruptedException {
 
        try {
            // Grab the Scheduler instance from the Factory
            Scheduler scheduler = StdSchedulerFactory.getDefaultScheduler();
 
            // and start it off
            scheduler.start();
 
            // define the job and tie it to our HelloJob class
            JobDetail job = JobBuilder.newJob(SelfDisAllowConExeTestJob.class)
                    .withIdentity("job1", "group1")
                    .build();
 
            // Trigger the job to run now, and then repeat every 40 seconds
            Trigger trigger = TriggerBuilder.newTrigger()
                    .withIdentity("trigger1", "group1")
                    .startNow()
                    .withSchedule(SimpleScheduleBuilder.simpleSchedule()
                            .withIntervalInSeconds(1)
                            .repeatForever())
                    .build();
 
            // Tell quartz to schedule the job using our trigger
            scheduler.scheduleJob(job, trigger);
 
            // wait trigger
            Thread.sleep(20000);
            scheduler.shutdown();
 
        } catch (SchedulerException se) {
            se.printStackTrace();
        }
    }
}

我们在实际代码中经常会结合spring，特地去看了一下MethodInvokingJobDetailFactoryBean的concurrent属性来控制是否限制并发执行的实现：

        Class<?> jobClass = (this.concurrent ? MethodInvokingJob.class : StatefulMethodInvokingJob.class);

    /**
     * Extension of the MethodInvokingJob, implementing the StatefulJob interface.
     * Quartz checks whether or not jobs are stateful and if so,
     * won't let jobs interfere with each other.
     */
    @PersistJobDataAfterExecution
    @DisallowConcurrentExecution
    public static class StatefulMethodInvokingJob extends MethodInvokingJob {
 
        // No implementation, just an addition of the tag interface StatefulJob
        // in order to allow stateful method invoking jobs.
    }

当然，在quartz里有一个StatefulJob，方便直接继承就实现了concurrent=false的事情了。

那么啰嗦了这么多，其实就是想表达，quartz里并没有一个可以设置说是否并发的接口，而是需要自定义的job自行继承，或使用注解来实现的。

另外，还有一个相关的注解：@PersistJobDataAfterExecution

意思是：放在JobDetail 里的JobDataMap是共享的，也就是相同任务之间执行时可以传输信息。很容易想到既然是共享的，那么就会有并发的问题，就如开头说的这个场景就会导致并发问题。所以官方文档也特别解释这个注解最好和@DisallowConcurrentExecution一起使用。

以下是例子：

@PersistJobDataAfterExecution
public class PersistJob implements Job {
    public void execute(JobExecutionContext context) throws JobExecutionException {
        JobDataMap data = context.getJobDetail().getJobDataMap();
        int i = data.getInt("P");
        System.out.printf("PersistJob=>"+i);
        i++;
        data.put("P", i);
    }
}

测试代码：

public class PersistJobDataQuartzTest {
    public static void main(String[] args) throws InterruptedException {
 
        try {
            // Grab the Scheduler instance from the Factory
            Scheduler scheduler = StdSchedulerFactory.getDefaultScheduler();
 
            // and start it off
            scheduler.start();
 
            JobDataMap jobDataMap = new JobDataMap();
            jobDataMap.put("P",1);
            // define the job and tie it to our HelloJob class
            JobDetail job = JobBuilder.newJob(PersistJob.class)
                    .withIdentity("job1", "group1").usingJobData(jobDataMap)
                    .build();
 
            // Trigger the job to run now, and then repeat every 40 seconds
            Trigger trigger = TriggerBuilder.newTrigger()
                    .withIdentity("trigger1", "group1")
                    .startNow()
                    .withSchedule(SimpleScheduleBuilder.simpleSchedule()
                            .withIntervalInSeconds(1)
                            .repeatForever())
                    .build();
 
            // Tell quartz to schedule the job using our trigger
            scheduler.scheduleJob(job, trigger);
            // wait trigger
            Thread.sleep(20000);
            scheduler.shutdown();
 
        } catch (SchedulerException se) {
            se.printStackTrace();
        }
    }
}

参考文档：

https://jayvilalta.com/blog/2014/06/04/understanding-the-disallowconcurrentexecution-job-attribute/

http://www.quartz-scheduler.org/documentation/quartz-2.1.x/tutorials/tutorial-lesson-03

http://www.cnblogs.com/lnlvinso/p/4194725.html