C#当中的多线程_线程池

3.1 简介

线程池主要用在需要大量短暂的开销大的资源的情形。我们预先分配一些资源在线程池当中,当我们需要使用的时候,直接从池中取出,代替了重新创建,不用时候就送回到池当中。

.NET当中的线程池是受CLR来管理的。

.NET线程池有一个QueueUserWorkItem()的静态方法,这个方法接收一个委托,每当该方法被调用后,委托进入内部的队列中,如果线程池当中没有任何线程,此时创建一个新的工作线程,并将队列的第一个委托放入到工作线程当中。

C#当中的多线程_线程池

注意点:

①线程池内的操作,尽量放入短时间运行的工作

②ASP.NET应用程序使用线程池不要把工作线程全部使用掉,否则web服务器将不能处理新的请求。

ASP.NET只推荐使用输入/输出密集型异步操作,因为其使用了一个不同的方式,叫做I/O线程。

③线程池当中的线程全部是后台线程,因此要注意前台线程执行完成后,后台线程也将结束工作。

3.2线程池中调用委托

首先要了解一个什么是【异步编程模型(Asynchronous Programming Model简称APM)】

.NET 1.0 异步编程模型(APM),

.NET 2.0 基于事件的异步编程模型(EAP),

.NET 4.0 基于任务的异步编程模型(TAP)。

本章主要了解什么是APM和EAP。

下面这篇文章介绍了异步编程模型,感觉挺好的,这里mark一下。

http://blog.csdn.net/xinke453/article/details/37810823

结合我这本书上的Demo,感觉理解起来无鸭梨,哈哈~

     class Program

     {

         static void Main(string[] args)

         {

             int threadId = ;

             RunOnThreadPool poolDelegate = Test;

                         //用创建线程的方法先创建了一个线程

             var t = new Thread(() => Test(out threadId));

             t.Start();

             t.Join();

             Console.WriteLine("Thread id: {0}", threadId);

             /*

16              使用BeginInvoke来运行委托,Callback是一个回调函数,

17               "a delegate asynchronous call" 代表你希望转发给回调方法的一个对象的引用,

18              在回调方法中,可以查询IAsyncResult接口的AsyncState属性来访问该对象

19              */

             IAsyncResult r = poolDelegate.BeginInvoke(out threadId, Callback, "a delegate asynchronous call");

              //这个例子当中使用AsyncWaitHandle属性来等待直到操作完成★

             r.AsyncWaitHandle.WaitOne();

             //操作完成后,会得到一个结果,可以通过委托调用EndInvoke方法,将IAsyncResult对象传递给委托参数。

             string result = poolDelegate.EndInvoke(out threadId, r);

             Console.WriteLine("Thread pool worker thread id: {0}", threadId);

             Console.WriteLine(result);

             Thread.Sleep(TimeSpan.FromSeconds());

         }

         private delegate string RunOnThreadPool(out int threadId);

         private static void Callback(IAsyncResult ar)

         {

             Console.WriteLine("Starting a callback...");

             Console.WriteLine("State passed to a callbak: {0}", ar.AsyncState);

             Console.WriteLine("Is thread pool thread: {0}", Thread.CurrentThread.IsThreadPoolThread);

             Console.WriteLine("Thread pool worker thread id: {0}", Thread.CurrentThread.ManagedThreadId);

         }

         private static string Test(out int threadId)

         {

             Console.WriteLine("Starting...");

             Console.WriteLine("Is thread pool thread: {0}", Thread.CurrentThread.IsThreadPoolThread);

             Thread.Sleep(TimeSpan.FromSeconds());

             threadId = Thread.CurrentThread.ManagedThreadId;

             return string.Format("Thread pool worker thread id was: {0}", threadId);

C#当中的多线程_线程池

注意:在这个例子当中,主线程调用Thread.Sleep(TimeSpan.FromSeconds(2));如果没这句话,回调函数就不会被执行了,

以为线程池是后台线程,此时主线程结束,那么后台线程也跟着结束了,所以可能不会执行。

对于访问异步操作的结果,APM提供了四种方式供开发人员选择:

①在调用BeginXxx方法的线程上调用EndXxx方法来得到异步操作的结果,但是这种方式会阻塞调用线程,知道操作完成之后调用线程才继续运行。

②查询IAsyncResult的AsyncWaitHandle属性,从而得到WaitHandle,然后再调用它的WaitOne方法来使一个线程阻塞并等待操作完成再调用EndXxx方法来获得操作的结果。

(本例子当中使用了这个方法)

③循环查询IAsyncResult的IsComplete属性,操作完成后再调用EndXxx方法来获得操作返回的结果。

④使用 AsyncCallback委托来指定操作完成时要调用的方法,在操作完成后调用的方法中调用EndXxx操作来获得异步操作的结果。

★★★推荐使用第④种方法,因为此时不会阻塞执行BeginXxx方法的线程,然而其他三种都会阻塞调用线程,相当于效果和使用同步方法是一样,个人感觉根本失去了异步编程的特点,所以其他三种方式可以简单了解下,在实际异步编程中都是使用委托的方式。

3.3向线程池中加入异步操作

QueueUserWorkItem方法的定义!
    [SecuritySafeCritical]
    public static bool QueueUserWorkItem(WaitCallback callBack);
[SecuritySafeCritical]
    public static bool QueueUserWorkItem(WaitCallback callBack, object state);
实例:
    class Program
    {
        static void Main(string[] args)
        {
            const int x = ;
            const int y = ;
            const string lambdaState = "lambda state 2";
           //方法一,直接调用QueueUserWorkItem传入单个参数,作为回调函数
           ThreadPool.QueueUserWorkItem(AsyncOperation);
          Thread.Sleep(TimeSpan.FromSeconds());           //方法二,传入回调函数以及状态参数
          ThreadPool.QueueUserWorkItem(AsyncOperation, "async state");
          Thread.Sleep(TimeSpan.FromSeconds());           //方法三,使用labmbda表达式
          ThreadPool.QueueUserWorkItem( state => {
                    Console.WriteLine("Operation state: {0}", state);
                    Console.WriteLine("Worker thread id: {0}", Thread.CurrentThread.ManagedThreadId);
                    Thread.Sleep(TimeSpan.FromSeconds());
                }, "lambda state");           //方法四,使用闭包机制
          ThreadPool.QueueUserWorkItem( _ =>
            {
                Console.WriteLine("Operation state: {0}, {1}", x+y, lambdaState);
                Console.WriteLine("Worker thread id: {0}", Thread.CurrentThread.ManagedThreadId);
                Thread.Sleep(TimeSpan.FromSeconds());
            }, "lambda state");             Thread.Sleep(TimeSpan.FromSeconds());
        }         private static void AsyncOperation(object state)
        {
            Console.WriteLine("Operation state: {0}", state ?? "(null)");
            Console.WriteLine("Worker thread id: {0}", Thread.CurrentThread.ManagedThreadId);
            Thread.Sleep(TimeSpan.FromSeconds());
        }

C#当中的多线程_线程池

扩展:C#闭包(Closure)机制是什么?

3.4线程池与并行度

下面这个实例展示线程池如何工作于大量异步操作,以及他和创建的大量单独线程方式的区别。

     class Program
    {
        static void Main(string[] args)
        {
            const int numberOfOperations = ;
            var sw = new Stopwatch();
            sw.Start();
            UseThreads(numberOfOperations);
            sw.Stop();
            Console.WriteLine("Execution time using threads: {0}", sw.ElapsedMilliseconds);             sw.Reset();
            sw.Start();
            UseThreadPool(numberOfOperations);
            sw.Stop();
            Console.WriteLine("Execution time using threads: {0}", sw.ElapsedMilliseconds);
        }         static void UseThreads(int numberOfOperations)
        {
            using (var countdown = new CountdownEvent(numberOfOperations))
            {
                Console.WriteLine("Scheduling work by creating threads");
                for (int i = ; i < numberOfOperations; i++)
                {
                    var thread = new Thread(() => {
                        Console.Write("{0},", Thread.CurrentThread.ManagedThreadId);
                        Thread.Sleep(TimeSpan.FromSeconds(0.1));
                        countdown.Signal();
                    });
                    thread.Start();
                }
                countdown.Wait();
                Console.WriteLine();
            }
        }         static void UseThreadPool(int numberOfOperations)
        {
            using (var countdown = new CountdownEvent(numberOfOperations))
            {
                Console.WriteLine("Starting work on a threadpool");
                for (int i = ; i < numberOfOperations; i++)
                {
                    ThreadPool.QueueUserWorkItem( _ => {
                        Console.Write("{0},", Thread.CurrentThread.ManagedThreadId);
                        Thread.Sleep(TimeSpan.FromSeconds(0.1));
                        countdown.Signal();
                    });
                }
                countdown.Wait();
                Console.WriteLine();
            }
        }
    }

C#当中的多线程_线程池

C#当中的多线程_线程池

分别用创建大量线程的方式和线程池的方式执行500个Thread.Sleep(TimeSpan.FromSeconds(0.1))操作,

我们发现线程池花费了更多的时间,但是占用的资源数目很少(通过ThreadId来看)。

3.5实现一个取消选项

使用CancellationTokenSource和CancellationToken两个类来实现工作线程工作的取消操作。

实例:
    class Program
    {
        static void Main(string[] args)
        {
             using (var cts = new CancellationTokenSource())
             {
                  CancellationToken token = cts.Token;
                  ThreadPool.QueueUserWorkItem(_ => AsyncOperation1(token));
                  Thread.Sleep(TimeSpan.FromSeconds());
                 cts.Cancel();
            }             using (var cts = new CancellationTokenSource())
            {
                 CancellationToken token = cts.Token;
                 ThreadPool.QueueUserWorkItem(_ => AsyncOperation2(token));
                 Thread.Sleep(TimeSpan.FromSeconds());
                 cts.Cancel();
             }              using (var cts = new CancellationTokenSource())
             {
                  CancellationToken token = cts.Token;
                 ThreadPool.QueueUserWorkItem(_ => AsyncOperation3(token));
                  Thread.Sleep(TimeSpan.FromSeconds());
                  cts.Cancel();
             }             Thread.Sleep(TimeSpan.FromSeconds());
        }        /// <summary>
      /// 第一种采用轮询IsCancellationRequested属性的方式,如果为true,那么操作被取消
       /// </summary>
       /// <param name="token"></param>
        static void AsyncOperation1(CancellationToken token)
        {
            Console.WriteLine("Starting the first task");
            for (int i = ; i < ; i++)
            {
                if (token.IsCancellationRequested)
                {
                    Console.WriteLine("The first task has been canceled.");
                    return;
                }
                Thread.Sleep(TimeSpan.FromSeconds());
            }
            Console.WriteLine("The first task has completed succesfully");
        }
       /// <summary>
       /// 抛出一个OperationCancelledException异常
       /// 这个允许操作之外控制取消过程,即需要取消操作的时候,通过操作之外的代码来处理
       /// </summary>
       /// <param name="token"></param>
        static void AsyncOperation2(CancellationToken token)
        {
            try
            {
                Console.WriteLine("Starting the second task");                 for (int i = ; i < ; i++)
                {
                    token.ThrowIfCancellationRequested();
                    Thread.Sleep(TimeSpan.FromSeconds());
                }
                Console.WriteLine("The second task has completed succesfully");
            }
            catch (OperationCanceledException)
            {
                Console.WriteLine("The second task has been canceled.");
            }
        }
       /// <summary>
       /// 第三种注册一个回调函数,当操作被取消时候,调用回调函数
       /// </summary>
       /// <param name="token"></param>
        private static void AsyncOperation3(CancellationToken token)
        {
            bool cancellationFlag = false;
            token.Register(() => cancellationFlag = true);
            Console.WriteLine("Starting the third task");
            for (int i = ; i < ; i++)
            {
                if (cancellationFlag)
                {
                    Console.WriteLine("The third task has been canceled.");
                    return;
                }
                Thread.Sleep(TimeSpan.FromSeconds());
            }
            Console.WriteLine("The third task has completed succesfully");
      

C#当中的多线程_线程池

CancellationTokenSource和CancellationToken两个类是.net4.0一会引入的,目前是实现异步操作取消事实标准。

3.6在线程池中使用等待事件处理器和超时

使用线程池当中的Threadpool.RegisterWaitSingleObject类来进行事件案处理。

RegisterWaitSingleObject的原型如下:

  public static RegisteredWaitHandle RegisterWaitForSingleObject(
          WaitHandle waitObject,
          WaitOrTimerCallback callBack,
          Object state,
        int millisecondsTimeOutInterval,
         bool executeOnlyOnce
)

参数

waitObject

要注册的 WaitHandle。使用 WaitHandle 而非 Mutex

callBack

waitObject 参数终止时调用的 WaitOrTimerCallback 委托。

state

传递给委托的对象。

timeout

TimeSpan 表示的超时时间。如果 timeout 为零,则函数测试对象的状态并立即返回。如果 timeout 为 -1,则函数的超时间隔永远不过期。

executeOnlyOnce

如果为 true,表示在调用了委托后,线程将不再在 waitObject 参数上等待;如果为 false,表示每次完成等待操作后都重置计时器,直到注销等待。

返回值

封装本机句柄的 RegisteredWaitHandle

相信看了这些之后大家还是一头雾水,这个方法的做用是向线程池添加一个可以定时执行的方法,第四个参数millisecondsTimeOutInterval 就是用来设置间隔执行的时间,但是这里第五个参数executeOnlyOnce 会对第四个参数起作用,当它为true时,表示任务仅会执行一次,就是说它不会,像Timer一样,每隔一定时间执行一次,这个功能的话用Timer控件也可以实现

该方法还在此基础上提供了基于信号量来触发执行任务。

信号量也叫开关量,故名思议,它只有两种状态,不是true就是false,

WaitHandle就是这类开关量的基础类,继承它的类有Mutex,ManualResetEvent,AutoResetEvent,一般我们使用后两个

写法:

static ManualResetEvent wait2=new ManualResetEvent(false);

static AutoResetEvent wait=new AutoResetEvent(false);

我们可以在将其实例化时指定开关量的初始值。(true为有信号,false为没信号)

ManualResetEvent和AutoResetEvent的区别在于:

前者调用Set方法后将自动将开关量值将一直保持为true,后者调用Set方法将变为true随后立即变为false,可以将它理解为一个脉冲。

例子
    class Program
    {
        static void Main(string[] args)
        {
            //执行两次RunOperations操作,第一次会超时,第二次不会超时
            RunOperations(TimeSpan.FromSeconds());
            RunOperations(TimeSpan.FromSeconds());
        }         static void RunOperations(TimeSpan workerOperationTimeout)
        {
            //定义一个ManualResetEvent信号量,初始为false
            using (var evt = new ManualResetEvent(false))
            //实例化一个CancellationTokenSource实例,用于取消操作
            using (var cts = new CancellationTokenSource())
            {
                Console.WriteLine("Registering timeout operations...");
                //注册超时的被调用的回调函数。
                var worker = ThreadPool.RegisterWaitForSingleObject(
                                        evt,
                    (state, isTimedOut) => WorkerOperationWait(cts, isTimedOut),
                                        null,
                                        workerOperationTimeout,
                                        true );                 Console.WriteLine("Starting long running operation...");
                //线程池执行WorkerOperation操作
                ThreadPool.QueueUserWorkItem(_ => WorkerOperation(cts.Token, evt));                 Thread.Sleep(workerOperationTimeout.Add(TimeSpan.FromSeconds()));
                worker.Unregister(evt);
            }
        }        /// <summary>
       /// 线程池内需要被调用的操作
       /// </summary>
       /// <param name="token"></param>
       /// <param name="evt"></param>
        static void WorkerOperation(CancellationToken token, ManualResetEvent evt)
        {
            for(int i = ; i < ; i++)
            {
                if (token.IsCancellationRequested)
                {
                    return;
                }
                Thread.Sleep(TimeSpan.FromSeconds());
            }
            //设置信号量,此时evt为true。
            evt.Set();
        }         /// <summary>
        /// 超时时候执行的回调函数
        /// </summary>
        /// <param name="cts"></param>
        /// <param name="isTimedOut"></param>
        static void WorkerOperationWait(CancellationTokenSource cts, bool isTimedOut)
        {
            if (isTimedOut)
            {
                cts.Cancel();
                Console.WriteLine("Worker operation timed out and was canceled.");
            }
            else
            {
                Console.WriteLine("Worker operation succeded.");
            }
     

C#当中的多线程_线程池

3.7在线程池中使用计时器

使用system.Threading.Timer对象来在线程池中创建周期性调用的异步操作。

     class Program
    {
        static void Main(string[] args)
        {
            Console.WriteLine("Press 'Enter' to stop the timer...");
            DateTime start = DateTime.Now;
            //实例化这个timer类
            //一秒后执行TimerOperation这个操作,然后每隔2秒执行一次
            _timer = new Timer(
                                _ => TimerOperation(start),
                                null,
                                TimeSpan.FromSeconds(),
                                TimeSpan.FromSeconds());             Thread.Sleep(TimeSpan.FromSeconds());             //改变计时器的运行时间,一秒后执行TimerOperation,然后每隔4秒执行一次
            _timer.Change(TimeSpan.FromSeconds(), TimeSpan.FromSeconds());             Console.ReadLine();             _timer.Dispose();
        }         static Timer _timer;         static void TimerOperation(DateTime start)
        {
            TimeSpan elapsed = DateTime.Now - start;
            Console.WriteLine("{0} seconds from {1}. Timer thread pool thread id: {2}", elapsed.Seconds, start,
            Thread.CurrentThread.ManagedThreadId);
        }

C#当中的多线程_线程池

3.8使用BackgroudWorker组件

本小节将介绍一个异步编程模式的另一种方式,叫基于事件的异步模式(EAP)

先看一个例子吧:

     class Program
    {
        static void Main(string[] args)
        {
              //实例化一个BackgroundWorker类
             var bw = new BackgroundWorker();
              //获取或设置一个值,该值指示 BackgroundWorker 能否报告进度更新。
             bw.WorkerReportsProgress = true;
              //设置后台工作线程是否支持取消操作
            bw.WorkerSupportsCancellation = true;             //给DoWork、ProgressChanged、RunWorkerCompleted事件绑定处理函数
            bw.DoWork += Worker_DoWork;
            bw.ProgressChanged += Worker_ProgressChanged;
            bw.RunWorkerCompleted += Worker_Completed;             //启动异步操作
            bw.RunWorkerAsync();             Console.WriteLine("Press C to cancel work");
            do
            {
                if (Console.ReadKey(true).KeyChar == 'C')
                {
                    //取消操作
                    bw.CancelAsync();
                }
                
            }
            while(bw.IsBusy);
        }         static void Worker_DoWork(object sender, DoWorkEventArgs e)
        {
            Console.WriteLine("DoWork thread pool thread id: {0}", Thread.CurrentThread.ManagedThreadId);
            var bw = (BackgroundWorker) sender;
            for (int i = ; i <= ; i++)
            {                 if (bw.CancellationPending)
                {
                    e.Cancel = true;
                    return;
                }                 if (i% == )
                {
                    bw.ReportProgress(i);
                }                 Thread.Sleep(TimeSpan.FromSeconds(0.1));
            }
            e.Result = ;
        }         static void Worker_ProgressChanged(object sender, ProgressChangedEventArgs e)
        {
            Console.WriteLine("{0}% completed. Progress thread pool thread id: {1}", e.ProgressPercentage,
                Thread.CurrentThread.ManagedThreadId);
        }         static void Worker_Completed(object sender, RunWorkerCompletedEventArgs e)
        {
            Console.WriteLine("Completed threadpool thread id:{0}",Thread.CurrentThread.ManagedThreadId);
            if (e.Error != null)
            {
                Console.WriteLine("Exception {0} has occured.", e.Error.Message);
            }
            else if (e.Cancelled)
            {
                Console.WriteLine("Operation has been canceled.");
            }
            else
            {
                Console.WriteLine("The answer is: {0}", e.Result);
            }
        }

事件

名称

说明

Disposed

当通过调用 Dispose 方法释放组件时发生。(从 Component 继承。)

DoWork

调用 RunWorkerAsync 时发生。

RunWorkerAsync 方法提交一个启动以异步方式运行的操作的请求。发出该请求后,将引发 DoWork 事件,该事件随后开始执行后台操作。

如果后台操作已在运行,则再次调用 RunWorkerAsync 将引发 InvalidOperationException

ProgressChanged

调用 ReportProgress 时发生。

public void ReportProgress

(
    int percentProgress
)

percentProgress

已完成的后台操作所占的百分比,范围从 0% 到 100%。

如果您需要后台操作报告其进度,则可以调用 ReportProgress 方法来引发 ProgressChanged 事件。 WorkerReportsProgress 属性值必须是
true,否则 ReportProgress 将引发 InvalidOperationException

您需要实现一个有意义的方法,以便按照占已完成的总任务的百分比来度量后台操作的进度。

对 ReportProgress 方法的调用为异步且立即返回。The ProgressChanged 事件处理程序在创建 BackgroundWorker 的线程上执行。

RunWorkerCompleted

当后台操作已完成、被取消或引发异常时发生。

在该方法中可以知道操作是成功完成还是发生错误,亦或被取消。

貼り付け元  <https://msdn.microsoft.com/zh-cn/library/system.componentmodel.backgroundworker.aspx>

成功完成的时候

C#当中的多线程_线程池

任务取消的时候

C#当中的多线程_线程池

上一篇:基于ELK的数据分析实践——满满的干货送给你


下一篇:Hadoop3.x完全分布式环境搭建