11、Java并发 Java CountDownLatch

本章节我们来讨论下 java.util.concurrent.CountDownLatch 这个类,顺带演示下如何在一些实际例子中使用它。

CountDownLatch 类的作用呢? 怎么说呢? 简单来说,我们可以使用它来阻塞线程,直到其他线程完成给定任务。

并发编程中使用 CountDownLatch

简而言之,CountDownLatch 有一个计数器字段,我们可以根据需要减少它,因此,我们可以使用它来阻止调用线程,直到它被计数到零。

如果我们正在进行一些并行处理,我们可以使用与计数器相同的值来实例化 CountDownLatch,因为我们想要处理多个线程。然后,我们可以在每个线程完成后调用 countdown(),保证调用 await() 的依赖线程将阻塞,直到工作线程完成。

使用 CountDownLatch 等待线程池完成

我们通过创建一个 Worker 来尝试这个模式,并使用 CountDownLatch 字段来指示它何时完成

public class Worker implements Runnable {
    private List<String> outputScraper;
    private CountDownLatch countDownLatch;

    public Worker(List<String> outputScraper, CountDownLatch countDownLatch) {
        this.outputScraper = outputScraper;
        this.countDownLatch = countDownLatch;
    }

    @Override
    public void run() {
        doSomeWork();
        outputScraper.add("Counted down");
        countDownLatch.countDown();
    }
}

然后,我们创建一个测试,以证明我们可以让 CountDownLatch 等待 Worker 实例完成

@Test
public void whenParallelProcessing_thenMainThreadWillBlockUntilCompletion()
  throws InterruptedException {

    List<String> outputScraper = Collections.synchronizedList(new ArrayList<>());
    CountDownLatch countDownLatch = new CountDownLatch(5);
    List<Thread> workers = Stream
      .generate(() -> new Thread(new Worker(outputScraper, countDownLatch)))
      .limit(5)
      .collect(toList());

      workers.forEach(Thread::start);
      countDownLatch.await(); 
      outputScraper.add("Latch released");

      assertThat(outputScraper)
        .containsExactly(
          "Counted down",
          "Counted down",
          "Counted down",
          "Counted down",
          "Counted down",
          "Latch released"
        );
}

上面这个示例中,”Latch release” 将始终是最后一个输出 – 因为它取决于 CountDownLatch 的释放。

注意,如果我们没有调用 await() 方法,我们将无法保证线程执行的顺序,因此测试会随机失败。

在等待开始的线程池中使用 CountDownLatch

我们重用前面的示例,但是这次开启了了数千个线程而不是 5 个线程,很可能许多早期的线程在后面的线程上调用 start() 之前已经完成了处理。这可能会使尝试重现并发问题变得困难,因为我们无法让所有线程并行运行。

为了解决这个问题,我们让 CountdownLatch 的工作方式与上一个示例有所不同。在某些子线程完成之前,我们可以阻止每个子线程直到所有其他子线程都启动,而不是阻塞父线程。

我们把上一个示例的 run() 方法修改下,使其在处理之前阻塞

public class WaitingWorker implements Runnable {

    private List<String> outputScraper;
    private CountDownLatch readyThreadCounter;
    private CountDownLatch callingThreadBlocker;
    private CountDownLatch completedThreadCounter;

    public WaitingWorker(
      List<String> outputScraper,
      CountDownLatch readyThreadCounter,
      CountDownLatch callingThreadBlocker,
      CountDownLatch completedThreadCounter) {

        this.outputScraper = outputScraper;
        this.readyThreadCounter = readyThreadCounter;
        this.callingThreadBlocker = callingThreadBlocker;
        this.completedThreadCounter = completedThreadCounter;
    }

    @Override
    public void run() {
        readyThreadCounter.countDown();
        try {
            callingThreadBlocker.await();
            doSomeWork();
            outputScraper.add("Counted down");
        } catch (InterruptedException e) {
            e.printStackTrace();
        } finally {
            completedThreadCounter.countDown();
        }
    }
}

接下来,我们修改下测试,直到所有工人都已启动,解锁工人,然后阻止,直到工人完成

@Test
public void whenDoingLotsOfThreadsInParallel_thenStartThemAtTheSameTime()
 throws InterruptedException {

    List<String> outputScraper = Collections.synchronizedList(new ArrayList<>());
    CountDownLatch readyThreadCounter = new CountDownLatch(5);
    CountDownLatch callingThreadBlocker = new CountDownLatch(1);
    CountDownLatch completedThreadCounter = new CountDownLatch(5);
    List<Thread> workers = Stream
      .generate(() -> new Thread(new WaitingWorker(
        outputScraper, readyThreadCounter, callingThreadBlocker, completedThreadCounter)))
      .limit(5)
      .collect(toList());

    workers.forEach(Thread::start);
    readyThreadCounter.await(); 
    outputScraper.add("Workers ready");
    callingThreadBlocker.countDown(); 
    completedThreadCounter.await(); 
    outputScraper.add("Workers complete");

    assertThat(outputScraper)
      .containsExactly(
        "Workers ready",
        "Counted down",
        "Counted down",
        "Counted down",
        "Counted down",
        "Counted down",
        "Workers complete"
      );
}

这种模式对于尝试重现并发错误非常有用,可以用来强制数千个线程尝试并行执行某些逻辑。

让 CountdownLatch 尽早结束

有时,我们可能会遇到一个情况,即在 CountdownLatch 倒计时之前,Workers 已经终止了错误。这可能导致它永远不会达到零并且 await() 永远不会终止。

@Override
public void run() {
    if (true) {
        throw new RuntimeException("Oh dear, I'm a BrokenWorker");
    }
    countDownLatch.countDown();
    outputScraper.add("Counted down");
}

我们修改下之前的测试以使用 BrokenWorker,来演示 await() 将如何永久阻塞

@Test
public void whenFailingToParallelProcess_thenMainThreadShouldGetNotGetStuck()
  throws InterruptedException {

    List<String> outputScraper = Collections.synchronizedList(new ArrayList<>());
    CountDownLatch countDownLatch = new CountDownLatch(5);
    List<Thread> workers = Stream
      .generate(() -> new Thread(new BrokenWorker(outputScraper, countDownLatch)))
      .limit(5)
      .collect(toList());

    workers.forEach(Thread::start);
    countDownLatch.await();
}

显然,这不是我们想要的行为 – 应用程序继续比无限阻塞要好得多。

为了解决这个问题,我们在调用 await() 时添加一个超时参数。

boolean completed = countDownLatch.await(3L, TimeUnit.SECONDS);
assertThat(completed).isFalse();

然后,我们可以看到,测试最终会超时,await() 将返回 false