ConcurrentHashMap在多线程应用中没有给出正确的输出

Question

我创建了一个例子来理解并发HashMap，但它似乎在7-8次运行后给出了不确定的输出。

在我的例子中，我创建了三个线程（它们模仿三个服务来获取数学，科学，英语的分数），它更新了一个只有3个键值对的共享HashMap（键是三个名字A，B，C并在运行结束时的值应该是三个科目的累积分数）

我发布了下面的代码，请注意。

错误的输出如下所示（正确的应该结束等来全部完成主：{C = 27，A = 57，B = 42}）

Within run Math : {C=0, A=0, B=0} 
Within run Science : {C=0, A=0, B=0} 
Completed Science : {C=10, A=39, B=29} 
Completed Math : {C=10, A=39, B=29} 
Within run English : {C=0, A=0, B=0} 
Completed English : {C=18, A=57, B=42} 
All Done main : {C=18, A=57, B=42} 

ConcurrentHashMapExample类：

package com.ll.thread.concurrency; 

    import java.util.Iterator; 
    import java.util.concurrent.ConcurrentHashMap; 
    import java.util.concurrent.CountDownLatch; 

    public class ConcurrentHashMapExample { 

     public static void main(String[] args) { 
      // TODO Auto-generated method stub 
      ConcurrentHashMap<String, Integer> concurrentHashMap = new ConcurrentHashMap<String, Integer>(); 
      concurrentHashMap.put("A", 0); 
      concurrentHashMap.put("B", 0); 
      concurrentHashMap.put("C", 0); 
      CountDownLatch countDownLatch = new CountDownLatch(3); 

      Runnable runnableScience = new Worker(concurrentHashMap , "Science" , countDownLatch); 
      Runnable runnableMath = new Worker(concurrentHashMap , "Math" , countDownLatch); 
      Runnable runnableEnglish = new Worker(concurrentHashMap , "English" , countDownLatch); 

      new Thread(runnableScience , "Science").start(); 
      new Thread(runnableMath ,"Math").start(); 
      new Thread(runnableEnglish ,"English").start(); 

      try { 
       countDownLatch.await(); 
      } catch (InterruptedException e) { 
       // TODO Auto-generated catch block 
       e.printStackTrace(); 
      } 
      System.out.println("All Done " + Thread.currentThread().getName() + " : "+concurrentHashMap); 




      concurrentHashMap = null; 



     } 

    } 

Worker类：​​

package com.ll.thread.concurrency; 

    import java.util.Iterator; 
    import java.util.concurrent.ConcurrentHashMap; 
    import java.util.concurrent.CountDownLatch; 

    public class Worker implements Runnable{ 

     ConcurrentHashMap<String, Integer> concurrentHashMap ; 
     String threadName = ""; 
     CountDownLatch countDownLatch ; 

     public Worker(ConcurrentHashMap<String, Integer> concurrentHashMap, 
       String subject, CountDownLatch countDownLatch) { 
      this.concurrentHashMap = concurrentHashMap; 
      this.threadName = subject ; 
      this.countDownLatch = countDownLatch; 

      // TODO Auto-generated constructor stub 
     } 

     @Override 
     public void run() { 
      // TODO Auto-generated method stub 
      System.out.println("Within run " +Thread.currentThread().getName() + " : "+concurrentHashMap); 
      try{ 
     for(Iterator<String> iterator = concurrentHashMap.keySet().iterator() ; iterator.hasNext();){ 
      String key = iterator.next(); 

      //synchronized (this) { 


      if("Math".equals(Thread.currentThread().getName())) { 
      if("A".equals(key)) 
       concurrentHashMap.put(key, concurrentHashMap.get(key) +20); 
      else if("B".equals(key)) 
       concurrentHashMap.put(key, concurrentHashMap.get(key) + 15); 
      else if("C".equals(key)) 
       concurrentHashMap.put(key, concurrentHashMap.get(key) +10); 
      }else 

      if("Science".equals(Thread.currentThread().getName())) { 
       if("A".equals(key)) 
        concurrentHashMap.put(key, concurrentHashMap.get(key) +19); 
       else if("B".equals(key)) 
        concurrentHashMap.put(key, concurrentHashMap.get(key) +14); 
       else if("C".equals(key)) 
        concurrentHashMap.put(key, concurrentHashMap.get(key) +9); 
       } 
      else 
       if("English".equals(Thread.currentThread().getName())) { 
        if("A".equals(key)) 
         concurrentHashMap.put(key, concurrentHashMap.get(key) +18); 
        else if("B".equals(key)) 
         concurrentHashMap.put(key, concurrentHashMap.get(key) +13); 
        else if("C".equals(key)) 
         concurrentHashMap.put(key, concurrentHashMap.get(key) +8); 
        } 

      } 
     // } 
      } 
      finally{ 

       System.out.println("Completed " + Thread.currentThread().getName() + " : " + concurrentHashMap); 
       countDownLatch.countDown(); 
      } 
     } 
     } 

Answer 1

我不认为你在你的测试需要ConcurrentHashMap。您在线程执行之前初始化映射。相反，你可以使用普通HashMap通过String键和AtomicInteger值参数化：

HashMap<String, AtomicInteger> concurrentHashMap = ... 
concurrentHashMap.put("A", new AtomicInteger(0)); 
... 

然后你就可以原子增加每个线程AtomicInteger实例的值：

concurrentHashMap.get(key).addAndGet(20);