ThreadPoolExecutor参数的用法及说明_Java教程

ThreadPoolExecutor参数说明

一、ThreadPoolExecutor核心参数说明

1、corePoolSize：核心线程数

* 核心线程会一直存活，及时没有任务需要执行

* 当线程数小于核心线程数时，即使有线程空闲，线程池也会优先创建新线程处理

* 设置allowCoreThreadTimeout=true（默认false）时，核心线程会超时关闭

2、queueCapacity：任务队列容量（阻塞队列）

* 当核心线程数达到最大时，新任务会放在队列中排队等待执行

3、maxPoolSize：最大线程数

* 当线程数>=corePoolSize，且任务队列已满时。线程池会创建新线程来处理任务

* 当线程数=maxPoolSize，且任务队列已满时，线程池会拒绝处理任务而抛出异常

4、 keepAliveTime：线程空闲时间

* 当线程空闲时间达到keepAliveTime时，线程会退出，直到线程数量=corePoolSize

* 如果allowCoreThreadTimeout=true，则会直到线程数量=0

5、allowCoreThreadTimeout：允许核心线程超时

6、rejectedExecutionHandler：任务拒绝处理器

* 两种情况会拒绝处理任务：

- 当线程数已经达到maxPoolSize，切队列已满，会拒绝新任务
- 当线程池被调用shutdown()后，会等待线程池里的任务执行完毕，再shutdown。如果在调用shutdown()和线程池真正shutdown之间提交任务，会拒绝新任务

* 线程池会调用rejectedExecutionHandler来处理这个任务。如果没有设置默认是AbortPolicy，会抛出异常

* ThreadPoolExecutor类有几个内部实现类来处理这类情况：

- AbortPolicy 丢弃任务，抛运行时异常
- CallerRunsPolicy 执行任务
- DiscardPolicy 忽视，什么都不会发生
- DiscardOldestPolicy 从队列中踢出最先进入队列（最后一个执行）的任务

* 实现RejectedExecutionHandler接口，可自定义处理器

二、ThreadPoolExecutor执行顺序

线程池按以下行为执行任务

1. 当线程数小于核心线程数时，创建线程。

2. 当线程数大于等于核心线程数，且任务队列未满时，将任务放入任务队列。

3. 当线程数大于等于核心线程数，且任务队列已满

-1 若线程数小于最大线程数，创建线程
-2 若线程数等于最大线程数，抛出异常，拒绝任务

三、ThreadPoolExecutor如何设置参数

1、默认值

* corePoolSize=1 * queueCapacity=Integer.MAX_VALUE * maxPoolSize=Integer.MAX_VALUE * keepAliveTime=60s * allowCoreThreadTimeout=false * rejectedExecutionHandler=AbortPolicy()

2、如何来设置

* 需要根据几个值来决定

- tasks ：每秒的任务数，假设为1000
- taskcost：每个任务花费时间，假设为0.1s
- responsetime：系统允许容忍的最大响应时间，假设为1s

* 做几个计算

- corePoolSize = 每秒需要多少个线程处理？

* 一颗CPU核心同一时刻只能执行一个线程，然后操作系统切换上下文，核心开始执行另一个线程的代码，以此类推，超过cpu核心数，就会放入队列，如果队列也满了，就另起一个新的线程执行，所有推荐：corePoolSize = ((cpu核心数 * 2) + 有效磁盘数)，java可以使用Runtime.getRuntime().availableProcessors()获取cpu核心数

- queueCapacity = (coreSizePool/taskcost)*responsetime

* 计算可得 queueCapacity = corePoolSize/0.1*1。意思是队列里的线程可以等待1s，超过了的需要新开线程来执行

* 切记不能设置为Integer.MAX_VALUE，这样队列会很大，线程数只会保持在corePoolSize大小，当任务陡增时，不能新开线程来执行，响应时间会随之陡增。

- maxPoolSize = (max(tasks)- queueCapacity)/(1/taskcost)

* 计算可得 maxPoolSize = (1000-corePoolSize)/10，即(每秒并发数-corePoolSize大小) / 10

* （最大任务数-队列容量）/每个线程每秒处理能力 = 最大线程数

- rejectedExecutionHandler：根据具体情况来决定，任务不重要可丢弃，任务重要则要利用一些缓冲机制来处理
- keepAliveTime和allowCoreThreadTimeout采用默认通常能满足

ThreadPoolExecutor参数allowCoreThreadTimeOut

ThreadPoolExecutor的执行流程有一点可能被吐槽过，就是只有缓存队列已经满了的时候才会使用到maxPoolSize创建新的线程.也就是说如果corePoolSize设为0的时候，要等到队列满了，才会创建线程去执行任务

之前有被问到，希望没有任务的时候线程池里的线程可以停掉。可能对性能和资源有过考虑的人都会想到这个问题吧

今天看JDK源码的时候发现了ThreadPoolExecutor在1.6的时候已经支持了

allowCoreThreadTimeOut参数就是为此设计的

				?

									/**

									 * Sets the policy governing whether core threads may time out and

									 * terminate if no tasks arrive within the keep-alive time, being

									 * replaced if needed when new tasks arrive. When false, core

									 * threads are never terminated due to lack of incoming

									 * tasks. When true, the same keep-alive policy applying to

									 * non-core threads applies also to core threads. To avoid

									 * continual thread replacement, the keep-alive time must be

									 * greater than zero when setting {@code true}. This method

									 * should in general be called before the pool is actively used.

									 *

									 * @param value {@code true} if should time out, else {@code false}

									 * @throws IllegalArgumentException if value is {@code true}

									 *         and the current keep-alive time is not greater than zero

									 *

									 * @since 1.6

									 */

									public void allowCoreThreadTimeOut(boolean value) {

									    if (value && keepAliveTime <= 0)

									        throw new IllegalArgumentException("Core threads must have nonzero keep alive times");

									    if (value != allowCoreThreadTimeOut) {

									        allowCoreThreadTimeOut = value;

									        if (value)

									            interruptIdleWorkers();

									    }

									}

在ThreadPoolExecutor构造函数的注释上也有明确说明：corePoolSize 的数量会一直保持，即使这些线程是空闲的，除非设置了allowCoreThreadTimeOut

				?

									/**

									     * Creates a new {@code ThreadPoolExecutor} with the given initial

									     * parameters and default thread factory and rejected execution handler.

									     * It may be more convenient to use one of the {@link Executors} factory

									     * methods instead of this general purpose constructor.

									     *

									     * @param corePoolSize the number of threads to keep in the pool, even

									     *        if they are idle, unless {@code allowCoreThreadTimeOut} is set

									     * @param maximumPoolSize the maximum number of threads to allow in the

									     *        pool

									     * @param keepAliveTime when the number of threads is greater than

									     *        the core, this is the maximum time that excess idle threads

									     *        will wait for new tasks before terminating.

									     * @param unit the time unit for the {@code keepAliveTime} argument

									     * @param workQueue the queue to use for holding tasks before they are

									     *        executed.  This queue will hold only the {@code Runnable}

									     *        tasks submitted by the {@code execute} method.

									     * @throws IllegalArgumentException if one of the following holds:<br>

									     *         {@code corePoolSize < 0}<br>

									     *         {@code keepAliveTime < 0}<br>

									     *         {@code maximumPoolSize <= 0}<br>

									     *         {@code maximumPoolSize < corePoolSize}

									     * @throws NullPointerException if {@code workQueue} is null

									     */

									    public ThreadPoolExecutor(int corePoolSize,

									                              int maximumPoolSize,

									                              long keepAliveTime,

									                              TimeUnit unit,

									                              BlockingQueue<Runnable> workQueue) {

									        this(corePoolSize, maximumPoolSize, keepAliveTime, unit, workQueue,

									             Executors.defaultThreadFactory(), defaultHandler);

									    }

当然，在Executors的静态工厂里面的newCachedThreadPool提供了另外一种思路

				?

									coreSize为0

									SynchronousQueue最多只能有一个任务在队列里面

也就是说这个线程池的任务会被立即分配一个线程去处理，如果没有空闲的线程会立即创建线程。

在空闲的时候，线程数量会减少直至为0，这一点倒是满足了要求。可是队列中最多只会缓存一个任务，当任务的处理速度慢于任务进入线程池的速度时，线程数量就会不断膨胀。如果maxPoolSize设置成一个比较小的数字时，可能就会有大量任务被拒绝策略处理。

所以正如注释中所说，newCachedThreadPool只适合于任务处理速度很快的场景下。比如做一些计算，不需要依赖其它服务

				?

									/**

									     * Creates a thread pool that creates new threads as needed, but

									     * will reuse previously constructed threads when they are

									     * available.  These pools will typically improve the performance

									     * of programs that execute many short-lived asynchronous tasks.

									     * Calls to {@code execute} will reuse previously constructed

									     * threads if available. If no existing thread is available, a new

									     * thread will be created and added to the pool. Threads that have

									     * not been used for sixty seconds are terminated and removed from

									     * the cache. Thus, a pool that remains idle for long enough will

									     * not consume any resources. Note that pools with similar

									     * properties but different details (for example, timeout parameters)

									     * may be created using {@link ThreadPoolExecutor} constructors.

									     *

									     * @return the newly created thread pool

									     */

									    public static ExecutorService newCachedThreadPool() {

									        return new ThreadPoolExecutor(0, Integer.MAX_VALUE,

									                                      60L, TimeUnit.SECONDS,

									                                      new SynchronousQueue<Runnable>());