自拍偷在线精品自拍偷,亚洲欧美中文日韩v在线观看不卡

定時(shí)任務(wù)實(shí)現(xiàn)原理詳解

開(kāi)發(fā) 前端
所謂最小堆方案,正如我們上面所說(shuō)的,每當(dāng)有新任務(wù)加入的時(shí)候,會(huì)把需要即將要執(zhí)行的任務(wù)排到前面,同時(shí)會(huì)有一個(gè)線程不斷的輪詢判斷,如果當(dāng)前某個(gè)任務(wù)已經(jīng)到達(dá)執(zhí)行時(shí)間點(diǎn),就會(huì)立即執(zhí)行,具體實(shí)現(xiàn)代表就是 JDK 中的 Timer 定時(shí)器!

一、摘要

在很多業(yè)務(wù)的系統(tǒng)中,我們常常需要定時(shí)的執(zhí)行一些任務(wù),例如定時(shí)發(fā)短信、定時(shí)變更數(shù)據(jù)、定時(shí)發(fā)起促銷活動(dòng)等等。

本文會(huì)重點(diǎn)分析下單機(jī)的定時(shí)任務(wù)實(shí)現(xiàn)原理以及優(yōu)缺點(diǎn),分布式框架的實(shí)現(xiàn)原理會(huì)在后續(xù)文章中進(jìn)行分析。

從單機(jī)角度,定時(shí)任務(wù)實(shí)現(xiàn)主要有以下 3 種方案:

  • while + sleep 組合
  • 最小堆實(shí)現(xiàn)
  • 時(shí)間輪實(shí)現(xiàn)

二、while+sleep組合

while+sleep 方案,簡(jiǎn)單的說(shuō),就是定義一個(gè)線程,然后 while 循環(huán),通過(guò) sleep 延遲時(shí)間來(lái)達(dá)到周期性調(diào)度任務(wù)。

簡(jiǎn)單示例如下:

public static void main(String[] args) {
    final long timeInterval = 5000;
    new Thread(new Runnable() {
        @Override
        public void run() {
            while (true) {
                System.out.println(Thread.currentThread().getName() + "每隔5秒執(zhí)行一次");
                try {
                    Thread.sleep(timeInterval);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        }
    }).start();
}

實(shí)現(xiàn)上非常簡(jiǎn)單,如果我們想在創(chuàng)建一個(gè)每隔3秒鐘執(zhí)行一次任務(wù),怎么辦呢?

同樣的,也可以在創(chuàng)建一個(gè)線程,然后間隔性的調(diào)度方法;但是如果創(chuàng)建了大量這種類型的線程,這個(gè)時(shí)候會(huì)發(fā)現(xiàn)大量的定時(shí)任務(wù)線程在調(diào)度切換時(shí)性能消耗會(huì)非常大,而且整體效率低!

面對(duì)這種在情況,大佬們也想到了,于是想出了用一個(gè)線程將所有的定時(shí)任務(wù)存起來(lái),事先排好序,按照一定的規(guī)則來(lái)調(diào)度,這樣不就可以極大的減少每個(gè)線程的切換消耗嗎?

正因此,JDK 中的 Timer 定時(shí)器由此誕生了!

三、最小堆實(shí)現(xiàn)

所謂最小堆方案,正如我們上面所說(shuō)的,每當(dāng)有新任務(wù)加入的時(shí)候,會(huì)把需要即將要執(zhí)行的任務(wù)排到前面,同時(shí)會(huì)有一個(gè)線程不斷的輪詢判斷,如果當(dāng)前某個(gè)任務(wù)已經(jīng)到達(dá)執(zhí)行時(shí)間點(diǎn),就會(huì)立即執(zhí)行,具體實(shí)現(xiàn)代表就是 JDK 中的 Timer 定時(shí)器!

3.1、Timer

首先我們來(lái)一個(gè)簡(jiǎn)單的 Timer 定時(shí)器例子

public static void main(String[] args) {
    Timer timer = new Timer();
    //每隔1秒調(diào)用一次
    timer.schedule(new TimerTask() {
        @Override
        public void run() {
            System.out.println("test1");
        }
    }, 1000, 1000);
    //每隔3秒調(diào)用一次
    timer.schedule(new TimerTask() {
        @Override
        public void run() {
            System.out.println("test2");
        }
    }, 3000, 3000);

}

實(shí)現(xiàn)上,好像跟我們上面介紹的 while+sleep 方案差不多,同樣也是起一個(gè)TimerTask線程任務(wù),只不過(guò)共用一個(gè)Timer調(diào)度器。

下面我們一起來(lái)打開(kāi)源碼看看里面到底有些啥!

  • 進(jìn)入Timer.schedule()方法

從方法上可以看出,這里主要做參數(shù)驗(yàn)證,其中TimerTask是一個(gè)線程任務(wù),delay表示延遲多久執(zhí)行(單位毫秒),period表示多久執(zhí)行一次(單位毫秒)

public void schedule(TimerTask task, long delay, long period) {
    if (delay < 0)
        throw new IllegalArgumentException("Negative delay.");
    if (period <= 0)
        throw new IllegalArgumentException("Non-positive period.");
    sched(task, System.currentTimeMillis()+delay, -period);
}
  • 接著看sched()方法

這步操作中,可以很清晰的看到,在同步代碼塊里,會(huì)將task對(duì)象加入到queue

private void sched(TimerTask task, long time, long period) {
    if (time < 0)
        throw new IllegalArgumentException("Illegal execution time.");

    // Constrain value of period sufficiently to prevent numeric
    // overflow while still being effectively infinitely large.
    if (Math.abs(period) > (Long.MAX_VALUE >> 1))
        period >>= 1;

    synchronized(queue) {
        if (!thread.newTasksMayBeScheduled)
            throw new IllegalStateException("Timer already cancelled.");

        synchronized(task.lock) {
            if (task.state != TimerTask.VIRGIN)
                throw new IllegalStateException(
                    "Task already scheduled or cancelled");
            task.nextExecutionTime = time;
            task.period = period;
            task.state = TimerTask.SCHEDULED;
        }

        queue.add(task);
        if (queue.getMin() == task)
            queue.notify();
    }
}
  • 我們繼續(xù)來(lái)看queue對(duì)象

任務(wù)會(huì)將入到TaskQueue隊(duì)列中,同時(shí)在Timer初始化階段會(huì)將TaskQueue作為參數(shù)傳入到TimerThread線程中,并且起到線程

public class Timer {
    
    private final TaskQueue queue = new TaskQueue();

    private final TimerThread thread = new TimerThread(queue);

    public Timer() {
        this("Timer-" + serialNumber());
    }

    public Timer(String name) {
        thread.setName(name);
        thread.start();
    }

    //...
}
  • 而TaskQueue其實(shí)是一個(gè)最小堆的數(shù)據(jù)實(shí)體類,源碼如下

每當(dāng)有新元素加入的時(shí)候,會(huì)對(duì)原來(lái)的數(shù)組進(jìn)行重排,會(huì)將即將要執(zhí)行的任務(wù)排在數(shù)組的前面

class TaskQueue {
    
    private TimerTask[] queue = new TimerTask[128];


    private int size = 0;

    void add(TimerTask task) {
        // Grow backing store if necessary
        if (size + 1 == queue.length)
            queue = Arrays.copyOf(queue, 2*queue.length);

        queue[++size] = task;
        fixUp(size);
    }

    private void fixUp(int k) {
        while (k > 1) {
            int j = k >> 1;
            if (queue[j].nextExecutionTime <= queue[k].nextExecutionTime)
                break;
            TimerTask tmp = queue[j];
            queue[j] = queue[k];
            queue[k] = tmp;
            k = j;
        }
    }
    
    //....
}
  • 最后我們來(lái)看看TimerThread

TimerThread其實(shí)就是一個(gè)任務(wù)調(diào)度線程,首先從TaskQueue里面獲取排在最前面的任務(wù),然后判斷它是否到達(dá)任務(wù)執(zhí)行時(shí)間點(diǎn),如果已到達(dá),就會(huì)立刻執(zhí)行任務(wù)

class TimerThread extends Thread {

    boolean newTasksMayBeScheduled = true;

    private TaskQueue queue;

    TimerThread(TaskQueue queue) {
        this.queue = queue;
    }

    public void run() {
        try {
            mainLoop();
        } finally {
            // Someone killed this Thread, behave as if Timer cancelled
            synchronized(queue) {
                newTasksMayBeScheduled = false;
                queue.clear();  // Eliminate obsolete references
            }
        }
    }

    /**
     * The main timer loop.  (See class comment.)
     */
    private void mainLoop() {
        while (true) {
            try {
                TimerTask task;
                boolean taskFired;
                synchronized(queue) {
                    // Wait for queue to become non-empty
                    while (queue.isEmpty() && newTasksMayBeScheduled)
                        queue.wait();
                    if (queue.isEmpty())
                        break; // Queue is empty and will forever remain; die

                    // Queue nonempty; look at first evt and do the right thing
                    long currentTime, executionTime;
                    task = queue.getMin();
                    synchronized(task.lock) {
                        if (task.state == TimerTask.CANCELLED) {
                            queue.removeMin();
                            continue;  // No action required, poll queue again
                        }
                        currentTime = System.currentTimeMillis();
                        executionTime = task.nextExecutionTime;
                        if (taskFired = (executionTime<=currentTime)) {
                            if (task.period == 0) { // Non-repeating, remove
                                queue.removeMin();
                                task.state = TimerTask.EXECUTED;
                            } else { // Repeating task, reschedule
                                queue.rescheduleMin(
                                  task.period<0 ? currentTime   - task.period
                                                : executionTime + task.period);
                            }
                        }
                    }
                    if (!taskFired) // Task hasn't yet fired; wait
                        queue.wait(executionTime - currentTime);
                }
                if (taskFired)  // Task fired; run it, holding no locks
                    task.run();
            } catch(InterruptedException e) {
            }
        }
    }
}

總結(jié)這個(gè)利用最小堆實(shí)現(xiàn)的方案,相比 while + sleep 方案,多了一個(gè)線程來(lái)管理所有的任務(wù),優(yōu)點(diǎn)就是減少了線程之間的性能開(kāi)銷,提升了執(zhí)行效率;但是同樣也帶來(lái)的了一些缺點(diǎn),整體的新加任務(wù)寫(xiě)入效率變成了 O(log(n))。

同時(shí),細(xì)心的發(fā)現(xiàn),這個(gè)方案還有以下幾個(gè)缺點(diǎn):

  • 串行阻塞:調(diào)度線程只有一個(gè),長(zhǎng)任務(wù)會(huì)阻塞短任務(wù)的執(zhí)行,例如,A任務(wù)跑了一分鐘,B任務(wù)至少需要等1分鐘才能跑
  • 容錯(cuò)能力差:沒(méi)有異常處理能力,一旦一個(gè)任務(wù)執(zhí)行故障,后續(xù)任務(wù)都無(wú)法執(zhí)行

3.2、ScheduledThreadPoolExecutor

鑒于 Timer 的上述缺陷,從 Java 5 開(kāi)始,推出了基于線程池設(shè)計(jì)的 ScheduledThreadPoolExecutor 。

圖片圖片

其設(shè)計(jì)思想是,每一個(gè)被調(diào)度的任務(wù)都會(huì)由線程池來(lái)管理執(zhí)行,因此任務(wù)是并發(fā)執(zhí)行的,相互之間不會(huì)受到干擾。需要注意的是,只有當(dāng)任務(wù)的執(zhí)行時(shí)間到來(lái)時(shí),ScheduledThreadPoolExecutor 才會(huì)真正啟動(dòng)一個(gè)線程,其余時(shí)間 ScheduledThreadPoolExecutor 都是在輪詢?nèi)蝿?wù)的狀態(tài)。

簡(jiǎn)單的使用示例:

public static void main(String[] args) {
    ScheduledThreadPoolExecutor executor = new ScheduledThreadPoolExecutor(3);
    //啟動(dòng)1秒之后,每隔1秒執(zhí)行一次
    executor.scheduleAtFixedRate((new Runnable() {
        @Override
        public void run() {
            System.out.println("test3");
        }
    }),1,1, TimeUnit.SECONDS);
    //啟動(dòng)1秒之后,每隔3秒執(zhí)行一次
    executor.scheduleAtFixedRate((new Runnable() {
        @Override
        public void run() {
            System.out.println("test4");
        }
    }),1,3, TimeUnit.SECONDS);
}

同樣的,我們首先打開(kāi)源碼,看看里面到底做了啥

  • 進(jìn)入scheduleAtFixedRate()方法

首先是校驗(yàn)基本參數(shù),然后將任務(wù)作為封裝到ScheduledFutureTask線程中,ScheduledFutureTask繼承自RunnableScheduledFuture,并作為參數(shù)調(diào)用delayedExecute()方法進(jìn)行預(yù)處理

public ScheduledFuture<?> scheduleAtFixedRate(Runnable command,
                                              long initialDelay,
                                              long period,
                                              TimeUnit unit) {
    if (command == null || unit == null)
        throw new NullPointerException();
    if (period <= 0)
        throw new IllegalArgumentException();
    ScheduledFutureTask<Void> sft =
        new ScheduledFutureTask<Void>(command,
                                      null,
                                      triggerTime(initialDelay, unit),
                                      unit.toNanos(period));
    RunnableScheduledFuture<Void> t = decorateTask(command, sft);
    sft.outerTask = t;
    delayedExecute(t);
    return t;
}
  • 繼續(xù)看delayedExecute()方法

可以很清晰的看到,當(dāng)線程池沒(méi)有關(guān)閉的時(shí)候,會(huì)通過(guò)super.getQueue().add(task)操作,將任務(wù)加入到隊(duì)列,同時(shí)調(diào)用ensurePrestart()方法做預(yù)處理

private void delayedExecute(RunnableScheduledFuture<?> task) {
    if (isShutdown())
        reject(task);
    else {
        super.getQueue().add(task);
        if (isShutdown() &&
            !canRunInCurrentRunState(task.isPeriodic()) &&
            remove(task))
            task.cancel(false);
        else
            //預(yù)處理
            ensurePrestart();
    }
}

其中super.getQueue()得到的是一個(gè)自定義的new DelayedWorkQueue()阻塞隊(duì)列,數(shù)據(jù)存儲(chǔ)方面也是一個(gè)最小堆結(jié)構(gòu)的隊(duì)列,這一點(diǎn)在初始化new ScheduledThreadPoolExecutor()的時(shí)候,可以看出!

public ScheduledThreadPoolExecutor(int corePoolSize) {
    super(corePoolSize, Integer.MAX_VALUE, 0, NANOSECONDS,
          new DelayedWorkQueue());
}

打開(kāi)源碼可以看到,DelayedWorkQueue其實(shí)是ScheduledThreadPoolExecutor中的一個(gè)靜態(tài)內(nèi)部類,在添加的時(shí)候,會(huì)將任務(wù)加入到RunnableScheduledFuture數(shù)組中,同時(shí)線程池中的Woker線程會(huì)通過(guò)調(diào)用任務(wù)隊(duì)列中的take()方法獲取對(duì)應(yīng)的ScheduledFutureTask線程任務(wù),接著執(zhí)行對(duì)應(yīng)的任務(wù)線程

static class DelayedWorkQueue extends AbstractQueue<Runnable>
        implements BlockingQueue<Runnable> {

    private static final int INITIAL_CAPACITY = 16;
    private RunnableScheduledFuture<?>[] queue =
        new RunnableScheduledFuture<?>[INITIAL_CAPACITY];
    private final ReentrantLock lock = new ReentrantLock();
    private int size = 0;   

    //....

    public boolean add(Runnable e) {
        return offer(e);
    }

    public boolean offer(Runnable x) {
        if (x == null)
            throw new NullPointerException();
        RunnableScheduledFuture<?> e = (RunnableScheduledFuture<?>)x;
        final ReentrantLock lock = this.lock;
        lock.lock();
        try {
            int i = size;
            if (i >= queue.length)
                grow();
            size = i + 1;
            if (i == 0) {
                queue[0] = e;
                setIndex(e, 0);
            } else {
                siftUp(i, e);
            }
            if (queue[0] == e) {
                leader = null;
                available.signal();
            }
        } finally {
            lock.unlock();
        }
        return true;
    }

    public RunnableScheduledFuture<?> take() throws InterruptedException {
        final ReentrantLock lock = this.lock;
        lock.lockInterruptibly();
        try {
            for (;;) {
                RunnableScheduledFuture<?> first = queue[0];
                if (first == null)
                    available.await();
                else {
                    long delay = first.getDelay(NANOSECONDS);
                    if (delay <= 0)
                        return finishPoll(first);
                    first = null; // don't retain ref while waiting
                    if (leader != null)
                        available.await();
                    else {
                        Thread thisThread = Thread.currentThread();
                        leader = thisThread;
                        try {
                            available.awaitNanos(delay);
                        } finally {
                            if (leader == thisThread)
                                leader = null;
                        }
                    }
                }
            }
        } finally {
            if (leader == null && queue[0] != null)
                available.signal();
            lock.unlock();
        }
    }
}
  • 回到我們最開(kāi)始說(shuō)到的ScheduledFutureTask任務(wù)線程類,最終執(zhí)行任務(wù)的其實(shí)就是它

ScheduledFutureTask任務(wù)線程,才是真正執(zhí)行任務(wù)的線程類,只是繞了一圈,做了很多包裝,run()方法就是真正執(zhí)行定時(shí)任務(wù)的方法。

private class ScheduledFutureTask<V>
            extends FutureTask<V> implements RunnableScheduledFuture<V> {

    /** Sequence number to break ties FIFO */
    private final long sequenceNumber;

    /** The time the task is enabled to execute in nanoTime units */
    private long time;

    /**
     * Period in nanoseconds for repeating tasks.  A positive
     * value indicates fixed-rate execution.  A negative value
     * indicates fixed-delay execution.  A value of 0 indicates a
     * non-repeating task.
     */
    private final long period;

    /** The actual task to be re-enqueued by reExecutePeriodic */
    RunnableScheduledFuture<V> outerTask = this;

    /**
     * Overrides FutureTask version so as to reset/requeue if periodic.
     */
    public void run() {
        boolean periodic = isPeriodic();
        if (!canRunInCurrentRunState(periodic))
            cancel(false);
        else if (!periodic)
            ScheduledFutureTask.super.run();
        else if (ScheduledFutureTask.super.runAndReset()) {
            setNextRunTime();
            reExecutePeriodic(outerTask);
        }
    }
    
    //...
}

3.3、小結(jié)

ScheduledExecutorService 相比 Timer 定時(shí)器,完美的解決上面說(shuō)到的 Timer 存在的兩個(gè)缺點(diǎn)!

在單體應(yīng)用里面,使用 ScheduledExecutorService 可以解決大部分需要使用定時(shí)任務(wù)的業(yè)務(wù)需求!

但是這是否意味著它是最佳的解決方案呢?

我們發(fā)現(xiàn)線程池中 ScheduledExecutorService 的排序容器跟 Timer 一樣,都是采用最小堆的存儲(chǔ)結(jié)構(gòu),新任務(wù)加入排序效率是O(log(n)),執(zhí)行取任務(wù)是O(1)。

這里的寫(xiě)入排序效率其實(shí)是有空間可提升的,有可能優(yōu)化到O(1)的時(shí)間復(fù)雜度,也就是我們下面要介紹的時(shí)間輪實(shí)現(xiàn)!

四、時(shí)間輪實(shí)現(xiàn)

所謂時(shí)間輪(RingBuffer)實(shí)現(xiàn),從數(shù)據(jù)結(jié)構(gòu)上看,簡(jiǎn)單的說(shuō)就是循環(huán)隊(duì)列,從名稱上看可能感覺(jué)很抽象。

它其實(shí)就是一個(gè)環(huán)形的數(shù)組,如圖所示,假設(shè)我們創(chuàng)建了一個(gè)長(zhǎng)度為 8 的時(shí)間輪。

插入、取值流程:

1.當(dāng)我們需要新建一個(gè) 1s 延時(shí)任務(wù)的時(shí)候,則只需要將它放到下標(biāo)為 1 的那個(gè)槽中,2、3、...、7也同樣如此。

2.而如果是新建一個(gè) 10s 的延時(shí)任務(wù),則需要將它放到下標(biāo)為 2 的槽中,但同時(shí)需要記錄它所對(duì)應(yīng)的圈數(shù),也就是 1 圈,不然就和 2 秒的延時(shí)消息重復(fù)了

3.當(dāng)創(chuàng)建一個(gè) 21s 的延時(shí)任務(wù)時(shí),它所在的位置就在下標(biāo)為 5 的槽中,同樣的需要為他加上圈數(shù)為 2,依次類推...

因此,總結(jié)起來(lái)有兩個(gè)核心的變量:

  • 數(shù)組下標(biāo):表示某個(gè)任務(wù)延遲時(shí)間,從數(shù)據(jù)操作上對(duì)執(zhí)行時(shí)間點(diǎn)進(jìn)行取余
  • 圈數(shù):表示需要循環(huán)圈數(shù)

通過(guò)這張圖可以更直觀的理解!

當(dāng)我們需要取出延時(shí)任務(wù)時(shí),只需要每秒往下移動(dòng)這個(gè)指針,然后取出該位置的所有任務(wù)即可,取任務(wù)的時(shí)間消耗為O(1)。

當(dāng)我們需要插入任務(wù)式,也只需要計(jì)算出對(duì)應(yīng)的下表和圈數(shù),即可將任務(wù)插入到對(duì)應(yīng)的數(shù)組位置中,插入任務(wù)的時(shí)間消耗為O(1)。

如果時(shí)間輪的槽比較少,會(huì)導(dǎo)致某一個(gè)槽上的任務(wù)非常多,那么效率也比較低,這就和 HashMap 的 hash 沖突是一樣的,因此在設(shè)計(jì)槽的時(shí)候不能太大也不能太小。

4.1、代碼實(shí)現(xiàn)

  • 首先創(chuàng)建一個(gè)RingBufferWheel時(shí)間輪定時(shí)任務(wù)管理器
public class RingBufferWheel {

    private Logger logger = LoggerFactory.getLogger(RingBufferWheel.class);


    /**
     * default ring buffer size
     */
    private static final int STATIC_RING_SIZE = 64;

    private Object[] ringBuffer;

    private int bufferSize;

    /**
     * business thread pool
     */
    private ExecutorService executorService;

    private volatile int size = 0;

    /***
     * task stop sign
     */
    private volatile boolean stop = false;

    /**
     * task start sign
     */
    private volatile AtomicBoolean start = new AtomicBoolean(false);

    /**
     * total tick times
     */
    private AtomicInteger tick = new AtomicInteger();

    private Lock lock = new ReentrantLock();
    private Condition condition = lock.newCondition();

    private AtomicInteger taskId = new AtomicInteger();
    private Map<Integer, Task> taskMap = new ConcurrentHashMap<>(16);

    /**
     * Create a new delay task ring buffer by default size
     *
     * @param executorService the business thread pool
     */
    public RingBufferWheel(ExecutorService executorService) {
        this.executorService = executorService;
        this.bufferSize = STATIC_RING_SIZE;
        this.ringBuffer = new Object[bufferSize];
    }


    /**
     * Create a new delay task ring buffer by custom buffer size
     *
     * @param executorService the business thread pool
     * @param bufferSize      custom buffer size
     */
    public RingBufferWheel(ExecutorService executorService, int bufferSize) {
        this(executorService);

        if (!powerOf2(bufferSize)) {
            throw new RuntimeException("bufferSize=[" + bufferSize + "] must be a power of 2");
        }
        this.bufferSize = bufferSize;
        this.ringBuffer = new Object[bufferSize];
    }

    /**
     * Add a task into the ring buffer(thread safe)
     *
     * @param task business task extends {@link Task}
     */
    public int addTask(Task task) {
        int key = task.getKey();
        int id;

        try {
            lock.lock();
            int index = mod(key, bufferSize);
            task.setIndex(index);
            Set<Task> tasks = get(index);

            int cycleNum = cycleNum(key, bufferSize);
            if (tasks != null) {
                task.setCycleNum(cycleNum);
                tasks.add(task);
            } else {
                task.setIndex(index);
                task.setCycleNum(cycleNum);
                Set<Task> sets = new HashSet<>();
                sets.add(task);
                put(key, sets);
            }
            id = taskId.incrementAndGet();
            task.setTaskId(id);
            taskMap.put(id, task);
            size++;
        } finally {
            lock.unlock();
        }

        start();

        return id;
    }


    /**
     * Cancel task by taskId
     * @param id unique id through {@link #addTask(Task)}
     * @return
     */
    public boolean cancel(int id) {

        boolean flag = false;
        Set<Task> tempTask = new HashSet<>();

        try {
            lock.lock();
            Task task = taskMap.get(id);
            if (task == null) {
                return false;
            }

            Set<Task> tasks = get(task.getIndex());
            for (Task tk : tasks) {
                if (tk.getKey() == task.getKey() && tk.getCycleNum() == task.getCycleNum()) {
                    size--;
                    flag = true;
                    taskMap.remove(id);
                } else {
                    tempTask.add(tk);
                }

            }
            //update origin data
            ringBuffer[task.getIndex()] = tempTask;
        } finally {
            lock.unlock();
        }

        return flag;
    }

    /**
     * Thread safe
     *
     * @return the size of ring buffer
     */
    public int taskSize() {
        return size;
    }

    /**
     * Same with method {@link #taskSize}
     * @return
     */
    public int taskMapSize(){
        return taskMap.size();
    }

    /**
     * Start background thread to consumer wheel timer, it will always run until you call method {@link #stop}
     */
    public void start() {
        if (!start.get()) {

            if (start.compareAndSet(start.get(), true)) {
                logger.info("Delay task is starting");
                Thread job = new Thread(new TriggerJob());
                job.setName("consumer RingBuffer thread");
                job.start();
                start.set(true);
            }

        }
    }

    /**
     * Stop consumer ring buffer thread
     *
     * @param force True will force close consumer thread and discard all pending tasks
     *              otherwise the consumer thread waits for all tasks to completes before closing.
     */
    public void stop(boolean force) {
        if (force) {
            logger.info("Delay task is forced stop");
            stop = true;
            executorService.shutdownNow();
        } else {
            logger.info("Delay task is stopping");
            if (taskSize() > 0) {
                try {
                    lock.lock();
                    condition.await();
                    stop = true;
                } catch (InterruptedException e) {
                    logger.error("InterruptedException", e);
                } finally {
                    lock.unlock();
                }
            }
            executorService.shutdown();
        }


    }


    private Set<Task> get(int index) {
        return (Set<Task>) ringBuffer[index];
    }

    private void put(int key, Set<Task> tasks) {
        int index = mod(key, bufferSize);
        ringBuffer[index] = tasks;
    }

    /**
     * Remove and get task list.
     * @param key
     * @return task list
     */
    private Set<Task> remove(int key) {
        Set<Task> tempTask = new HashSet<>();
        Set<Task> result = new HashSet<>();

        Set<Task> tasks = (Set<Task>) ringBuffer[key];
        if (tasks == null) {
            return result;
        }

        for (Task task : tasks) {
            if (task.getCycleNum() == 0) {
                result.add(task);

                size2Notify();
            } else {
                // decrement 1 cycle number and update origin data
                task.setCycleNum(task.getCycleNum() - 1);
                tempTask.add(task);
            }
            // remove task, and free the memory.
            taskMap.remove(task.getTaskId());
        }

        //update origin data
        ringBuffer[key] = tempTask;

        return result;
    }

    private void size2Notify() {
        try {
            lock.lock();
            size--;
            if (size == 0) {
                condition.signal();
            }
        } finally {
            lock.unlock();
        }
    }

    private boolean powerOf2(int target) {
        if (target < 0) {
            return false;
        }
        int value = target & (target - 1);
        if (value != 0) {
            return false;
        }

        return true;
    }

    private int mod(int target, int mod) {
        // equals target % mod
        target = target + tick.get();
        return target & (mod - 1);
    }

    private int cycleNum(int target, int mod) {
        //equals target/mod
        return target >> Integer.bitCount(mod - 1);
    }

    /**
     * An abstract class used to implement business.
     */
    public abstract static class Task extends Thread {

        private int index;

        private int cycleNum;

        private int key;

        /**
         * The unique ID of the task
         */
        private int taskId ;

        @Override
        public void run() {
        }

        public int getKey() {
            return key;
        }

        /**
         *
         * @param key Delay time(seconds)
         */
        public void setKey(int key) {
            this.key = key;
        }

        public int getCycleNum() {
            return cycleNum;
        }

        private void setCycleNum(int cycleNum) {
            this.cycleNum = cycleNum;
        }

        public int getIndex() {
            return index;
        }

        private void setIndex(int index) {
            this.index = index;
        }

        public int getTaskId() {
            return taskId;
        }

        public void setTaskId(int taskId) {
            this.taskId = taskId;
        }
    }


    private class TriggerJob implements Runnable {

        @Override
        public void run() {
            int index = 0;
            while (!stop) {
                try {
                    Set<Task> tasks = remove(index);
                    for (Task task : tasks) {
                        executorService.submit(task);
                    }

                    if (++index > bufferSize - 1) {
                        index = 0;
                    }

                    //Total tick number of records
                    tick.incrementAndGet();
                    TimeUnit.SECONDS.sleep(1);

                } catch (Exception e) {
                    logger.error("Exception", e);
                }

            }

            logger.info("Delay task has stopped");
        }
    }
}
  • 接著,編寫(xiě)一個(gè)客戶端,測(cè)試客戶端
public static void main(String[] args) {
    RingBufferWheel ringBufferWheel = new RingBufferWheel( Executors.newFixedThreadPool(2));
    for (int i = 0; i < 3; i++) {
        RingBufferWheel.Task job = new Job();
        job.setKey(i);
        ringBufferWheel.addTask(job);
    }
}

public static class Job extends RingBufferWheel.Task{
    @Override
    public void run() {
        System.out.println("test5");
    }
}

運(yùn)行結(jié)果:

test5
test5
test5

如果要周期性執(zhí)行任務(wù),可以在任務(wù)執(zhí)行完成之后,再重新加入到時(shí)間輪中。

詳細(xì)源碼分析地址:點(diǎn)擊這里獲取

4.2、應(yīng)用

時(shí)間輪的應(yīng)用還是非常廣的,例如在 Disruptor 項(xiàng)目中就運(yùn)用到了 RingBuffer,還有Netty中的HashedWheelTimer工具原理也差不多等等,有興趣的同學(xué),可以閱讀一下官方對(duì)應(yīng)的源碼!

五、小結(jié)

本文主要圍繞單體應(yīng)用中的定時(shí)任務(wù)原理進(jìn)行分析,可能也有理解不對(duì)的地方,歡迎評(píng)論區(qū)留言!

六、參考

1、https://www.jianshu.com/p/84d9db1b1def

2、https://crossoverjie.top/2019/09/27/algorithm

責(zé)任編輯:武曉燕 來(lái)源: 潘志的研發(fā)筆記
相關(guān)推薦

2020-12-21 07:31:23

實(shí)現(xiàn)單機(jī)JDK

2024-05-31 13:07:29

.NET Core定時(shí)任務(wù)編程

2021-06-30 07:19:34

SpringBoot定時(shí)任務(wù)

2024-09-20 05:49:04

SpringBoot后端

2024-10-15 16:41:35

2024-11-04 16:01:01

2024-07-17 17:54:28

XXLJob分布式任務(wù)Java

2019-12-25 15:10:00

MySQL事件數(shù)據(jù)庫(kù)

2023-12-19 08:09:06

Python定時(shí)任務(wù)Cron表達(dá)式

2020-12-10 07:00:38

編程線程池定時(shí)任務(wù)

2024-05-13 09:49:30

.NETQuartz庫(kù)Cron表達(dá)式

2017-08-16 16:41:04

JavaSpringBoot定時(shí)任務(wù)

2024-01-22 08:53:00

策略任務(wù)RocketMQ

2021-11-22 12:35:40

Python命令定時(shí)任務(wù)

2024-02-26 11:12:33

定時(shí)任務(wù)線程

2024-01-31 08:38:57

Python定時(shí)任務(wù)函數(shù)

2009-10-28 10:05:29

Ubuntucrontab定時(shí)任務(wù)

2012-02-07 13:31:14

SpringJava

2010-03-10 15:47:58

crontab定時(shí)任務(wù)

2024-02-28 09:54:07

線程池配置
點(diǎn)贊
收藏

51CTO技術(shù)棧公眾號(hào)