Java 的4个线程池

Java的并发包里面，Executors下面

• newCachedThreadPool
• newFixedThreadPool(5)
• newScheduledThreadPool(3)

• newSingleThreadExecutor

  /**
   * 线程池的类型
   * 线程池的作用
   * 线程池的原理
   */
  fun runCachedPool(){
      runPool("cached",10,Executors.newCachedThreadPool())
  }
  
  fun runFixedPool(){
      runPool("fixed",10,Executors.newFixedThreadPool(5))
  }
  
  fun runSchedulePool(){
      runPool("schedule",10,Executors.newScheduledThreadPool(3))
  }
  
  fun runSinglePool(){
      runPool("Single",10,Executors.newSingleThreadExecutor())
  }
  
  private fun runPool(method:String, nums:Int, executor: ExecutorService){
      println("run $method start")
      val countDownLatch = CountDownLatch(nums)
      for(i in 0 until nums){
          executor.submit(ThreadClz("$method ${i+1}",countDownLatch))
      }
      executor.shutdown()
      countDownLatch.await()
      println("run $method end")
  }
  
  class ThreadClz(var name:String, private var countDownLatch: CountDownLatch):Runnable{
  
      override fun run() {
          println("id:${Thread.currentThread().id}_$name running,time:${toYYMMDDHHmmssSSS()} start @${Thread.currentThread().name}")
          Thread.sleep(2000)
          println("id:${Thread.currentThread().id}_$name running,time:${toYYMMDDHHmmssSSS()} end")
          countDownLatch.countDown()
      }
  
      @SuppressLint("SimpleDateFormat", "WeekBasedYear")
      fun toYYMMDDHHmmssSSS():String{
          val date = Date()
          val sdf = SimpleDateFormat("YYYY/MM/dd HH:mm:ss.SSS")
          return sdf.format(date)
      }
  
  }

  newCachedThreadPool

  fun runCachedPool(){
      runPool("cached",10,Executors.newCachedThreadPool())
  }

  run cached start
  id:20_cached 1 running,time:2021/01/14 17:56:23.868 start @pool-3-thread-1
  id:21_cached 2 running,time:2021/01/14 17:56:23.868 start @pool-3-thread-2
  id:22_cached 3 running,time:2021/01/14 17:56:23.868 start @pool-3-thread-3
  id:23_cached 4 running,time:2021/01/14 17:56:23.868 start @pool-3-thread-4
  id:25_cached 6 running,time:2021/01/14 17:56:23.869 start @pool-3-thread-6
  id:24_cached 5 running,time:2021/01/14 17:56:23.869 start @pool-3-thread-5
  id:26_cached 7 running,time:2021/01/14 17:56:23.869 start @pool-3-thread-7
  id:27_cached 8 running,time:2021/01/14 17:56:23.869 start @pool-3-thread-8
  id:28_cached 9 running,time:2021/01/14 17:56:23.870 start @pool-3-thread-9
  id:29_cached 10 running,time:2021/01/14 17:56:23.870 start @pool-3-thread-10
  id:21_cached 2 running,time:2021/01/14 17:56:25.870 end
  id:20_cached 1 running,time:2021/01/14 17:56:25.870 end
  id:22_cached 3 running,time:2021/01/14 17:56:25.870 end
  id:26_cached 7 running,time:2021/01/14 17:56:25.873 end
  id:24_cached 5 running,time:2021/01/14 17:56:25.873 end
  id:25_cached 6 running,time:2021/01/14 17:56:25.873 end
  id:23_cached 4 running,time:2021/01/14 17:56:25.873 end
  id:27_cached 8 running,time:2021/01/14 17:56:25.875 end
  id:28_cached 9 running,time:2021/01/14 17:56:25.875 end
  id:29_cached 10 running,time:2021/01/14 17:56:25.875 end
  run cached end

  10个thread同时start，没有看到线程池的影子。。。
  newCachedThreadPool,不限制线程的个数。怎么复用？

  fun runCachedPool(){
      val method = "cached"
      val nums = 10
      val executor = Executors.newCachedThreadPool()
      println("run $method start")
      val countDownLatch = CountDownLatch(nums)
      for(i in 0 until nums/2){
          executor.execute(ThreadClz("$method ${i+1}",countDownLatch))
      }
      Thread.sleep(3000)
      for(i in nums/2 until nums){
          executor.execute(ThreadClz("$method ${i+1}",countDownLatch))
      }
  //        executor.shutdown()
      countDownLatch.await()
      println("run $method end")
  }

  结果如下：

  run cached start
  id:11_cached 1 running,time:2021/01/15 11:54:36.696 start @pool-1-thread-1
  id:14_cached 4 running,time:2021/01/15 11:54:36.696 start @pool-1-thread-4
  id:12_cached 2 running,time:2021/01/15 11:54:36.696 start @pool-1-thread-2
  id:13_cached 3 running,time:2021/01/15 11:54:36.696 start @pool-1-thread-3
  id:15_cached 5 running,time:2021/01/15 11:54:36.696 start @pool-1-thread-5
  id:15_cached 5 running,time:2021/01/15 11:54:38.713 end
  id:14_cached 4 running,time:2021/01/15 11:54:38.713 end
  id:11_cached 1 running,time:2021/01/15 11:54:38.713 end
  id:12_cached 2 running,time:2021/01/15 11:54:38.713 end
  id:13_cached 3 running,time:2021/01/15 11:54:38.713 end
  id:15_cached 8 running,time:2021/01/15 11:54:39.698 start @pool-1-thread-5
  id:11_cached 9 running,time:2021/01/15 11:54:39.698 start @pool-1-thread-1
  id:14_cached 10 running,time:2021/01/15 11:54:39.698 start @pool-1-thread-4
  id:12_cached 7 running,time:2021/01/15 11:54:39.698 start @pool-1-thread-2
  id:13_cached 6 running,time:2021/01/15 11:54:39.698 start @pool-1-thread-3
  id:15_cached 8 running,time:2021/01/15 11:54:41.703 end
  id:11_cached 9 running,time:2021/01/15 11:54:41.703 end
  id:12_cached 7 running,time:2021/01/15 11:54:41.703 end
  id:14_cached 10 running,time:2021/01/15 11:54:41.703 end
  id:13_cached 6 running,time:2021/01/15 11:54:41.703 end
  run cached end

  可以看到，后面5个线程被复用了。所以，cache的线程会保存60s，看来是真的。
  其他的三个，newFixedThreadPool，newScheduledThreadPool
  和newSingleThreadExecutor
  Sinle很简单，就是线程池只有1个线程在跑，其他task都在排队，等待线程空闲。
  其他2个都是会需要输入核心线程的个数，也就是线程池会持有几个线程。

先说结论：

• newCachedThreadPool，线程数不固定，线程空闲以后，会被回收。
• newFixedThreadPool, 核心线程数固定，最大线程数也是固定。核心线程的概念就是处于IDLE状态，也不会回收。
• newScheduledThreadPool，核心线程数固定，线程数无上限。它还可以进行delay操作，延迟进行线程运行。
• newSingleThreadExecutor，保证每个时刻只有1个线程在运行。single可以保证唯一性。线程回收以后，还能重新new一个线程。