Quartz是運用最廣的任務排程框架，它最核心的組成部分是Scheduler、Trigger、JobDetail，然後給Scheduler配置個執行緒QuartzSchedulerThread，此執行緒在Scheduler初始化時啟動，等待Scheduler start，然後從JobStore裡拿到最近要觸發的Trigger，以執行緒等待的方式等到trigger觸發時間點，之後就是執行trigger所關聯的JobDetail，最後清掃戰場。Scheduler初始化、start和trigger執行的時序圖如下所示：

其中，最核心的地方是QuartzSchedulerThread執行機制。下面解析一下它的run方法：

public void run() { boolean lastAcquireFailed = false; while (!halted) { try { // check if we're supposed to pause... synchronized (pauseLock) { while (paused && !halted) { try { // wait until togglePause(false) is called... pauseLock.wait(100L); } catch (InterruptedException ignore) { } } if (halted) { break; } } ...... } }

 以上是run的最開頭的一段，不難看出這是在等待scheduler的start，實際上Quartz就是通過執行緒的wait或sleep來實現時間排程。繼續看程式碼：

Trigger trigger = null; long now = System.currentTimeMillis(); signaled = false; try { trigger = qsRsrcs.getJobStore().acquireNextTrigger( ctxt, now + idleWaitTime); lastAcquireFailed = false; } catch (JobPersistenceException jpe) { if(!lastAcquireFailed) { qs.notifySchedulerListenersError( "An error occured while scanning for the next trigger to fire.", jpe); } lastAcquireFailed = true; } catch (RuntimeException e) { if(!lastAcquireFailed) { getLog().error("quartzSchedulerThreadLoop: RuntimeException " +e.getMessage(), e); } lastAcquireFailed = true; }

這段程式碼是從jobStore裡拿到下一個要執行的trigger，一般情況下jobStore使用的是RAMJobStore，即trigger等相關資訊存放在記憶體裡，如果需要把任務持久化就得使用可持久化JobStore。繼續看程式碼：

 now = System.currentTimeMillis(); long triggerTime = trigger.getNextFireTime().getTime(); long timeUntilTrigger = triggerTime - now; long spinInterval = 10; int numPauses = (int) (timeUntilTrigger / spinInterval); while (numPauses >= 0 && !signaled) { try { Thread.sleep(spinInterval); } catch (InterruptedException ignore) { } now = System.currentTimeMillis(); timeUntilTrigger = triggerTime - now; numPauses = (int) (timeUntilTrigger / spinInterval); } if (signaled) { try { qsRsrcs.getJobStore().releaseAcquiredTrigger( ctxt, trigger); } catch (JobPersistenceException jpe) { qs.notifySchedulerListenersError( "An error occured while releasing trigger '" + trigger.getFullName() + "'", jpe); // db connection must have failed... keep // retrying until it's up... releaseTriggerRetryLoop(trigger); } catch (RuntimeException e) { getLog().error( "releaseTriggerRetryLoop: RuntimeException " +e.getMessage(), e); // db connection must have failed... keep // retrying until it's up... releaseTriggerRetryLoop(trigger); } signaled = false; continue; }

此段程式碼是計算下一個trigger的執行時間和現在系統時間的差，然後通過迴圈執行緒sleep的方式暫停住此執行緒，一直等到trigger的執行時間點。繼續看程式碼：

import org.quartz.core.JobRunShell; JobRunShell shell = null; try { shell = qsRsrcs.getJobRunShellFactory().borrowJobRunShell(); shell.initialize(qs, bndle); } catch (SchedulerException se) { try { qsRsrcs.getJobStore().triggeredJobComplete(ctxt, trigger, bndle.getJobDetail(), Trigger.INSTRUCTION_SET_ALL_JOB_TRIGGERS_ERROR); } catch (SchedulerException se2) { qs.notifySchedulerListenersError( "An error occured while placing job's triggers in error state '" + trigger.getFullName() + "'", se2); // db connection must have failed... keep retrying // until it's up... errorTriggerRetryLoop(bndle); } continue; } if (qsRsrcs.getThreadPool().runInThread(shell) == false) { try { getLog().error("ThreadPool.runInThread() return false!"); qsRsrcs.getJobStore().triggeredJobComplete(ctxt, trigger, bndle.getJobDetail(), Trigger.INSTRUCTION_SET_ALL_JOB_TRIGGERS_ERROR); } catch (SchedulerException se2) { qs.notifySchedulerListenersError( "An error occured while placing job's triggers in error state '" + trigger.getFullName() + "'", se2); // db connection must have failed... keep retrying // until it's up... releaseTriggerRetryLoop(trigger); } }

此段程式碼就是包裝trigger，然後通過以JobRunShell為載體，在threadpool裡執行trigger所關聯的jobDetail。

之後的程式碼就是清掃戰場，就不在累述。