Java 线程池的拒绝策略

Java 线程池的拒绝策略

Java 线程池的拒绝策略一共有四种，分别为AbortPolicy，CallerRunsPolicy，DiscardPolicy，DiscardOldestPolicy。下面就来说明这四种策略的不同之处。

下面策略都需要使用 DemoTask 类，该类的代码如下：

package org.gettingreal.juc.learning.executors;

public class DemoTask implements Runnable {

    private int index;

    public DemoTask(int index) {
        this.index = index;
    }

    @Override
    public void run() {
        System.out.println("index: " + index + "，invoke thread: " + Thread.currentThread().getName());

        try {
            // simulate business code.
            Thread.sleep(2000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

}

AbortPolicy 策略

该策略丢弃任务，并抛出 RejectedExecutionException 异常信息。是线程池默认的拒绝策略。

package org.gettingreal.juc.learning.executors;

import java.util.concurrent.*;

/**
 * AbortPolicy 策略测试
 */
public class AbortPolicyTest {

    public static void main(String[] args) {
        int corePoolSize = 1;
        int maximumPoolSize = 3;
        long keepAliveTime = 60L;
        BlockingQueue<Runnable> workQueue = new LinkedBlockingQueue<Runnable>(2);

        RejectedExecutionHandler rejectedExecutionHandler = new ThreadPoolExecutor.AbortPolicy();

        ThreadPoolExecutor executor = new ThreadPoolExecutor(corePoolSize, maximumPoolSize, keepAliveTime,
                TimeUnit.MILLISECONDS, workQueue, rejectedExecutionHandler);

        for (int i = 0; i < 10; i++) {
            executor.submit(new DemoTask(i));
        }

        executor.shutdown();
    }

}

执行上述代码得到如下输出：

# 输出
index: 0，invoke thread: pool-1-thread-1
index: 4，invoke thread: pool-1-thread-3
index: 3，invoke thread: pool-1-thread-2
Exception in thread "main" java.util.concurrent.RejectedExecutionException: Task java.util.concurrent.FutureTask@29453f44 rejected from java.util.concurrent.ThreadPoolExecutor@5cad8086[Running, pool size = 3, active threads = 3, queued tasks = 2, completed tasks = 0]
	at java.util.concurrent.ThreadPoolExecutor$AbortPolicy.rejectedExecution(ThreadPoolExecutor.java:2063)
	at java.util.concurrent.ThreadPoolExecutor.reject(ThreadPoolExecutor.java:830)
	at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1379)
	at java.util.concurrent.AbstractExecutorService.submit(AbstractExecutorService.java:112)
	at org.gettingreal.juc.learning.executors.AbortPolicyTest.main(AbortPolicyTest.java:22)
index: 1，invoke thread: pool-1-thread-2
index: 2，invoke thread: pool-1-thread-3

可以看出，只有 5 个任务得到了执行，实际数量就是 maximumPoolSize + workQueue.size() 的值。

CallerRunsPolicy 策略

当触发拒绝策略，只要线程池没有关闭的话，则使用调用线程直接运行任务。一般对于并发比较小，性能要求不高，不允许失败的场景比较适用。但是，由于调用者自己运行任务，如果任务提交速度过快，可能导致程序阻塞，性能堪忧。

package org.gettingreal.juc.learning.executors;

import java.util.concurrent.*;

/**
 * CallerRunsPolicy 策略测试
 */
public class CallerRunsPolicyTest {

    public static void main(String[] args) {
        int corePoolSize = 1;
        int maximumPoolSize = 3;
        long keepAliveTime = 60L;
        BlockingQueue<Runnable> workQueue = new LinkedBlockingQueue<Runnable>(2);

        RejectedExecutionHandler rejectedExecutionHandler = new ThreadPoolExecutor.CallerRunsPolicy();

        ThreadPoolExecutor executor = new ThreadPoolExecutor(corePoolSize, maximumPoolSize, keepAliveTime,
                TimeUnit.MILLISECONDS, workQueue, rejectedExecutionHandler);

        for (int i = 0; i < 10; i++) {
            executor.submit(new DemoTask(i));
        }

        executor.shutdown();
    }

}

执行上面代码，得到如下输出：

# 输出
index: 0，invoke thread: pool-1-thread-1
index: 4，invoke thread: pool-1-thread-3
index: 5，invoke thread: main
index: 3，invoke thread: pool-1-thread-2
index: 6，invoke thread: main
index: 2，invoke thread: pool-1-thread-3
index: 1，invoke thread: pool-1-thread-1
index: 7，invoke thread: pool-1-thread-3
index: 8，invoke thread: pool-1-thread-1
index: 9，invoke thread: pool-1-thread-4

可以看到里面 main 线程执行相关的任务，所有的任务都得到了执行，并且没有抛出异常。

DiscardPolicy 策略

这个策略是直接丢弃。

package org.gettingreal.juc.learning.executors;

import java.util.concurrent.*;

/**
 * DiscardPolicy 策略测试
 */
public class DiscardPolicyTest {

    public static void main(String[] args) {
        int corePoolSize = 1;
        int maximumPoolSize = 3;
        long keepAliveTime = 60L;
        BlockingQueue<Runnable> workQueue = new LinkedBlockingQueue<Runnable>(2);

        RejectedExecutionHandler rejectedExecutionHandler = new ThreadPoolExecutor.DiscardPolicy();

        ThreadPoolExecutor executor = new ThreadPoolExecutor(corePoolSize, maximumPoolSize, keepAliveTime,
                TimeUnit.MILLISECONDS, workQueue, rejectedExecutionHandler);

        for (int i = 0; i < 10; i++) {
            executor.submit(new DemoTask(i));
        }

        executor.shutdown();
    }

}

执行上述代码，得到如下输出：

# 输出
index: 0，invoke thread: pool-1-thread-1
index: 3，invoke thread: pool-1-thread-2
index: 4，invoke thread: pool-1-thread-3
index: 1，invoke thread: pool-1-thread-3
index: 2，invoke thread: pool-1-thread-1

可以看出，有 5 个任务被拒绝了，也没有出现错误。

DiscardOldestPolicy 策略

当触发拒绝策略，只要线程池没有关闭的话，丢弃阻塞队列 workQueue 中最老的一个任务，并将新任务加入。

package org.gettingreal.juc.learning.executors;

import java.util.concurrent.*;

public class DiscardOldestPolicyTest {

    public static void main(String[] args) {
        int corePoolSize = 1;
        int maximumPoolSize = 3;
        long keepAliveTime = 60L;
        BlockingQueue<Runnable> workQueue = new LinkedBlockingQueue<Runnable>(2);

        RejectedExecutionHandler rejectedExecutionHandler = new ThreadPoolExecutor.DiscardOldestPolicy();

        ThreadPoolExecutor executor = new ThreadPoolExecutor(corePoolSize, maximumPoolSize, keepAliveTime,
                TimeUnit.MILLISECONDS, workQueue, rejectedExecutionHandler);

        for (int i = 0; i < 10; i++) {
            executor.submit(new DemoTask(i));
        }

        executor.shutdown();
    }
}

执行如上代码，得到如下输出：

index: 0，invoke thread: pool-1-thread-1
index: 4，invoke thread: pool-1-thread-3
index: 3，invoke thread: pool-1-thread-2
index: 8，invoke thread: pool-1-thread-2
index: 9，invoke thread: pool-1-thread-1

可以看出，index 后面的编号是最新加入线程池的任务。