Use of thread pool technology

Use of thread pool technology

1. Create a thread pool (not used)

ExecutorService threadPool1 = Executors.newFixedThreadPool (3); // fixed size
ExecutorService threadPool2 = Executors.newCachedThreadPool (); // elastically stretchable thread pool, strong, strong case ExecutorService threadPool3 = Executors.newSingleThreadExecutor (); // only one
which There are disadvantages of the thread pool created by this method (Ali's original words)
1) FixedThreadPool and SingleThreadPool: The
allowed request queue length is Integer.MAX_VALUE, which may accumulate a large number of requests, resulting in OOM.
2) CachedThreadPool and ScheduledThreadPool:
The number of allowed creation threads is Integer.MAX_VALUE, which may create a large number of threads, resulting in OOM.

The underlying creation methods are the same:

Executors.newFixedThreadPool(3)底层为：
 return new **ThreadPoolExecutor**(nThreads, nThreads,
                                      0L, TimeUnit.MILLISECONDS,
                                      new LinkedBlockingQueue<Runnable>());
                                      
Executors.newCachedThreadPool()底层为：
                                      return new **ThreadPoolExecutor**(0, Integer.MAX_VALUE,
                                      60L, TimeUnit.SECONDS,
                                      new SynchronousQueue<Runnable>());
                                      
Executors.newSingleThreadExecutor()底层为：
												new **ThreadPoolExecutor**(1, 1,
                                    0L, TimeUnit.MILLISECONDS,
                                    new LinkedBlockingQueue<Runnable>())

LinkedBlockingQueue does not specify the size, the default is the largest integer.

2. Recommended creation method (recommended)

// Get the current computer or server thread number,

    public class ThreadPoolTec {
    public static void main(String[] args) {
        //推荐的线程池的创建方法
        /**
         * 最多可以存在的人，maximumPoolSize + LinkedBlockingDeque的容量
         *触发非活跃线程变为活跃的方法是：当前任务数量大于最大线程池容量，每超一个则激活一个，直到都被激活
         * // 拒绝策略说明：
         * // 1. AbortPolicy （默认的：队列满了，就丢弃任务抛出异常！）
         * // 2. CallerRunsPolicy（哪来的回哪去？ 谁叫你来的，你就去哪里处理）
         * // 3. DiscardOldestPolicy (尝试将最早进入对立与的人任务删除,尝试加入队列)
         * // 4. DiscardPolicy (队列满了任务也会丢弃,不抛出异常)
         */
      Integer processors = Runtime.getRuntime().availableProcessors());
        ExecutorService threadPool = new ThreadPoolExecutor(
                2,//核心活跃线程数，类比银行两个柜台一直保持营业
                processors ,//线程池最大大小，类比银行共5个柜台可以营业
                2L,//超时回收空闲的线程，类比有三个非活跃线程处于活跃状态，在一定时间还未接到任务就进入非活跃状态（就是不营业了）
                TimeUnit.SECONDS,//时间单位
                new LinkedBlockingDeque<>(3),//存放等待任务的队列，类比为银行的候客区，不指定大小的话就是最大整数
                Executors.defaultThreadFactory(),// 线程工厂，不修改！用来创建
                new ThreadPoolExecutor.CallerRunsPolicy() // 拒绝策略，即候客区满了，不再允许其他人排队了

        );

        try {
            for(int j=1;j<=8;j++){
                threadPool.execute(()->{
                    System.out.println(Thread.currentThread().getName()+"is running………………");
                });
            }
        } catch (Exception e) {
            e.printStackTrace();
        }finally {
            threadPool.shutdown();
        }
    }
}

Here are a few notes:
1. The maximum number of threads to create a thread pool is the same as the number of threads running the machine is the maximum utilization of resources.
Runtime.getRuntime (). AvailableProcessors ()); is to obtain the maximum number of threads of the current machine
2. The queue of the task queue must specify the size, new LinkedBlockingDeque <> (3), so that there will not be too much waiting.
3. Judgment strategy depends on the specific requirements, I use new ThreadPoolExecutor.CallerRunsPolicy ()
4. Pay attention to the execution of multi-threads, written in lamda:
for (int j = 1; j <= 8; j ++) {
threadPool.execute ( ()-> {
System.out.println (Thread.currentThread (). GetName () + "is running ..................");
});
}

ThreadPoolExecutor underlying working principle:
Rejection strategy flow chart: