Quartz任务调度：MisFire策略和源码分析

2019-08-16 11:09:29来源：博客园阅读 ()

Quartz任务调度：MisFire策略和源码分析

Quartz是为大家熟知的任务调度框架，先看看官网的介绍：

-------------------------------------------------------------------------------------------------------------------------

What is the Quartz Job Scheduling Library?

Quartz is a richly featured, open source job scheduling library that can be integrated within virtually any Java application - from the smallest stand-alone application to the largest e-commerce system. Quartz can be used to create simple or complex schedules for executing tens, hundreds, or even tens-of-thousands of jobs; jobs whose tasks are defined as standard Java components that may execute virtually anything you may program them to do. The Quartz Scheduler includes many enterprise-class features, such as support for JTA transactions and clustering.

Quartz is freely usable, licensed under the Apache 2.0 license.

-------------------------------------------------------------------------------------------------------------------------

翻译：Quartz是一个功能丰富、开源的任务调度库，它可以集成到几乎任意Java应用中---小到最小的独立应用，大到最大的电子商务系统。Quartz 可以用来创建简单或者复杂的工作计划，同时执行数十、成百、甚至上万的任务。可被定义为标准Java组件的任务，几乎可以执行任意可以编程的任务。Quartz 任务调度包含许多企业级功能特性，比如支持JTA事务和集群。

Quartz可以免费试用，遵循 Apache 2.0 license 许可协议

-------------------------------------------------------------------------------------------------------------------------

公司项目也用的Quartz，最近遇到一些关于Quartz的问题，带着疑问，查阅了部分Quartz源码，与大家分享。

开始是为了研究Quartz的MisFire策略，当任务执行时间过长、服务停机、任务暂停等原因，导致其超过其下次执行的时间点时，就会涉及MisFire（失火，错误任务的触发）处理的策略问题。 Quartz的任务分为SimpleTrigger和CronTrigger，项目中一般使用CronTrigger居多，本文只涉及了CronTrigger的MisFire处理策略（SimpleTrigger的MisFire策略与CronTrigger不同，后续再说）。

MisFire策略常量的定义在类CronTrigger中，列举如下：

MISFIRE_INSTRUCTION_FIRE_ONCE_NOW = 1
MISFIRE_INSTRUCTION_DO_NOTHING = 2
MISFIRE_INSTRUCTION_SMART_POLICY = 0
MISFIRE_INSTRUCTION_IGNORE_MISFIRE_POLICY = -1

根据JavaDoc介绍和官网文档分析，其对应执行策略如下：

MISFIRE_INSTRUCTION_FIRE_ONCE_NOW：立即执行一次，然后按照Cron定义时间点执行
MISFIRE_INSTRUCTION_DO_NOTHING：什么都不做，等待Cron定义下次任务执行的时间点
MISFIRE_INSTRUCTION_SMART_POLICY：智能的策略，针对不同的Trigger执行不同，CronTrigger时为MISFIRE_INSTRUCTION_FIRE_ONCE_NOW
MISFIRE_INSTRUCTION_IGNORE_MISFIRE_POLICY：将所有错过的执行时间点全都补上，例如，任务15s执行一次，执行的任务错过了4分钟，则执行MisFire时，一次性执行4*(60/15)= 16次任务

但是，我写了例子，实际执行策略1和策略2与文档又不太相同，示例任务cron表达式为：0/10 * * * * ?，每10s执行一次。

测试步骤如下：

任务下次执行时间为15:05:10，misFire策略为MISFIRE_INSTRUCTION_FIRE_ONCE_NOW（1）

1.将任务暂停至15:05:35

2.重新启动任务，任务瞬间执行了3次

将misFire策略设置为MISFIRE_INSTRUCTION_DO_NOTHING与上述表现一致。这个实验结果与文档描述不太相符。

于是，翻阅Quartz源码，首先从定时任务本身入手，打断点，找到任务执行工作线程为：WorkerThread对象，工作线程池为：SimpleThreadPool

核心代码如下：

       // WorkerThread.class
       // 将任务送入工作线程
　　　　public void run(Runnable newRunnable) {
            synchronized(lock) {
                if(runnable != null) {
                    throw new IllegalStateException("Already running a Runnable!");
                }

                runnable = newRunnable;
                lock.notifyAll();
            }
        }
        //循环执行，当有任务送入时执行任务
        @Override
        public void run() {
            boolean ran = false;
            
            while (run.get()) {
                try {
                    synchronized(lock) {
                        while (runnable == null && run.get()) {
                            lock.wait(500);
                        }

                        if (runnable != null) {
                            ran = true;
                            runnable.run();
                        }
                    }
                } catch (InterruptedException unblock) {
                    // do nothing (loop will terminate if shutdown() was called
                    try {
                        getLog().error("Worker thread was interrupt()'ed.", unblock);
                    } catch(Exception e) {
                        // ignore to help with a tomcat glitch
                    }
                } catch (Throwable exceptionInRunnable) {
                    try {
                        getLog().error("Error while executing the Runnable: ",
                            exceptionInRunnable);
                    } catch(Exception e) {
                        // ignore to help with a tomcat glitch
                    }
                } finally {
                    synchronized(lock) {
                        runnable = null;
                    }
                    // repair the thread in case the runnable mucked it up...
                    if(getPriority() != tp.getThreadPriority()) {
                        setPriority(tp.getThreadPriority());
                    }

                    if (runOnce) {
                           run.set(false);
                        clearFromBusyWorkersList(this);
                    } else if(ran) {
                        ran = false;
                        makeAvailable(this);
                    }

                }
            }

可以看到当有任务送入工作线程时，任务将被执行。由此，反向找到线程池代码，代码如下：

    
　　　　// SimpleThreadPool.class
　　　　public boolean runInThread(Runnable runnable) {
        if (runnable == null) {
            return false;
        }

        synchronized (nextRunnableLock) {

            handoffPending = true;

            // Wait until a worker thread is available
            while ((availWorkers.size() < 1) && !isShutdown) {
                try {
                    nextRunnableLock.wait(500);
                } catch (InterruptedException ignore) {
                }
            }

            if (!isShutdown) {
                WorkerThread wt = (WorkerThread)availWorkers.removeFirst();
                busyWorkers.add(wt);
                wt.run(runnable);
            } else {
                // If the thread pool is going down, execute the Runnable
                // within a new additional worker thread (no thread from the pool).
                WorkerThread wt = new WorkerThread(this, threadGroup,
                        "WorkerThread-LastJob", prio, isMakeThreadsDaemons(), runnable);
                busyWorkers.add(wt);
                workers.add(wt);
                wt.start();
            }
            nextRunnableLock.notifyAll();
            handoffPending = false;
        }

        return true;
    }

可以看到线程池从可用的工作线程队列中取出一个工作线程，将任务送入工作线程（WorkerThread），然后任务会被执行。

由此，反向找到调用方法runInThread的地方，类QuartzSchedulerThread（约398行），QuartzSchedulerThread集成自Thread，又是一个无限循环执行的线程任务，找到类QuartzSchedulerThread.run()方法(由于代码量较大，此处不再全部粘贴)，可以看到这个方法干的活大概是：循环找出需要执行的Job，然后送入线程池，再由线程池送入工作线程。

列举部分关键代码：

1.找出需要执行的Job的代码

                    try {
　　　　　　　　　　　　　　//此处去数据库查询将要执行的任务
                        triggers = qsRsrcs.getJobStore().acquireNextTriggers(
                                now + idleWaitTime, Math.min(availThreadCount, qsRsrcs.getMaxBatchSize()), qsRsrcs.getBatchTimeWindow());
                        acquiresFailed = 0;
                        if (log.isDebugEnabled())
                            log.debug("batch acquisition of " + (triggers == null ? 0 : triggers.size()) + " triggers");
                    } catch (JobPersistenceException jpe) {
                        if (acquiresFailed == 0) {
                            qs.notifySchedulerListenersError(
                                "An error occurred while scanning for the next triggers to fire.",
                                jpe);
                        }
                        if (acquiresFailed < Integer.MAX_VALUE)
                            acquiresFailed++;
                        continue;
                    } catch (RuntimeException e) {
                        if (acquiresFailed == 0) {
                            getLog().error("quartzSchedulerThreadLoop: RuntimeException "
                                    +e.getMessage(), e);
                        }
                        if (acquiresFailed < Integer.MAX_VALUE)
                            acquiresFailed++;
                        continue;
                    }

关键点在注释处的代码，方法：acquireNextTriggers，继续debug跟进该方法，找到查询SQL，代码如下：

　　 // StdJDBCDelegate.class
    public List<TriggerKey> selectTriggerToAcquire(Connection conn, long noLaterThan, long noEarlierThan, int maxCount)
        throws SQLException {
        PreparedStatement ps = null;
        ResultSet rs = null;
        List<TriggerKey> nextTriggers = new LinkedList<TriggerKey>();
        try {
            ps = conn.prepareStatement(rtp(SELECT_NEXT_TRIGGER_TO_ACQUIRE));
            
            // Set max rows to retrieve
            if (maxCount < 1)
                maxCount = 1; // we want at least one trigger back.
            ps.setMaxRows(maxCount);
            
            // Try to give jdbc driver a hint to hopefully not pull over more than the few rows we actually need.
            // Note: in some jdbc drivers, such as MySQL, you must set maxRows before fetchSize, or you get exception!
            ps.setFetchSize(maxCount);
            
            ps.setString(1, STATE_WAITING);
            ps.setBigDecimal(2, new BigDecimal(String.valueOf(noLaterThan)));
            ps.setBigDecimal(3, new BigDecimal(String.valueOf(noEarlierThan)));
            rs = ps.executeQuery();
            
            while (rs.next() && nextTriggers.size() <= maxCount) {
                nextTriggers.add(triggerKey(
                        rs.getString(COL_TRIGGER_NAME),
                        rs.getString(COL_TRIGGER_GROUP)));
            }
            
            return nextTriggers;
        } finally {
            closeResultSet(rs);
            closeStatement(ps);
        }      
    }

根据debug时实时参数，处理过的SQL为：

SELECT
    TRIGGER_NAME,
    TRIGGER_GROUP,
    NEXT_FIRE_TIME,
    PRIORITY
FROM
    qrtz_TRIGGERS
WHERE
    SCHED_NAME = 'schedulerFactoryBean'
AND TRIGGER_STATE = 'WAITING'
AND NEXT_FIRE_TIME <= (now + idleWaitTime)
AND (
    MISFIRE_INSTR = -1
    OR (
        MISFIRE_INSTR != -1
        AND NEXT_FIRE_TIME >= (now - misfireThreshold)
    )
)
ORDER BY  NEXT_FIRE_TIME ASC, PRIORITY DESC

其中：now为系统当前时间，idleWaitTime为系统线程闲置时间，默认取值为30s，misfireThreshold为配置参数，意思为系统能容忍的misFire的最大阀值，默认为60s（当前系统配置也是60s，之前一直不知道这个值什么意思）。从SQL中看得很清楚了，这个SQL语句是要查询出：未来30s内将要执行的任务，且MISFIRE_INSTR为-1（MISFIRE_INSTRUCTION_IGNORE_MISFIRE_POLICY），或者MISFIRE_INSTR不为-1，但是，NEXT_FIRE_TIME错过的执行时间不能超过阀值60s。至此问题搞清楚了，影响misFire执行策略的另一个参数就是misfireThreshold，配置文件quartz.properties中，对应org.quartz.jobStore.misfireThreshold: 60000，单位毫秒。也就是说：如果【错过时间】不超过60s都不算是misFire，不执行misFire策略，依次执行错过的任务时间点；【错过时间】超过60s按misFire策略执行。

根据上述结论重新进行试验，将任务暂停时间超过60s，这次试验结果与文档描述一致。

另外，跟踪启动任务的代码，找到处理misFire的方法，代码位置：org.quartz.impl.triggers.CronTriggerImpl.updateAfterMisfire(Calendar)

    @Override
    public void updateAfterMisfire(org.quartz.Calendar cal) {
        int instr = getMisfireInstruction();

        if(instr == Trigger.MISFIRE_INSTRUCTION_IGNORE_MISFIRE_POLICY)
            return;

        if (instr == MISFIRE_INSTRUCTION_SMART_POLICY) {
            instr = MISFIRE_INSTRUCTION_FIRE_ONCE_NOW;
        }

        if (instr == MISFIRE_INSTRUCTION_DO_NOTHING) {
            Date newFireTime = getFireTimeAfter(new Date());
            while (newFireTime != null && cal != null
                    && !cal.isTimeIncluded(newFireTime.getTime())) {
                newFireTime = getFireTimeAfter(newFireTime);
            }
            setNextFireTime(newFireTime);
        } else if (instr == MISFIRE_INSTRUCTION_FIRE_ONCE_NOW) {
            setNextFireTime(new Date());
        }
    }