We Keep Coding

Handle exceptions with TPL Dataflow blocks

I have a simple tpl data flow which basically does some tasks.
I noticed when there is an exception in any of the datablocks, it wasn't getting caught in the initial parent block caller.
I have added some manual code to check for exception but doesn't seem the right approach.
if (readBlock.Completion.Exception != null
|| saveBlockJoinedProcess.Completion.Exception != null
|| processBlock1.Completion.Exception != null
|| processBlock2.Completion.Exception != null)
{
throw readBlock.Completion.Exception;
}
I had a look online to see what's a suggested approach but didn't see anything obvious.
So I created some sample code below and was hoping to get some guidance on a better solution:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading;
using System.Threading.Tasks;
using System.Threading.Tasks.Dataflow;
namespace TPLDataflow
{
class Program
{
static void Main(string[] args)
{
try
{
//ProcessB();
ProcessA();
}
catch (Exception e)
{
Console.WriteLine("Exception in Process!");
throw new Exception($"exception:{e}");
}
Console.WriteLine("Processing complete!");
Console.ReadLine();
}
private static void ProcessB()
{
Task.WhenAll(Task.Run(() => DoSomething(1, "ProcessB"))).Wait();
}
private static void ProcessA()
{
var random = new Random();
var readBlock = new TransformBlock<int, int>(x =>
{
try { return DoSomething(x, "readBlock"); }
catch (Exception e) { throw e; }
}); //1
var braodcastBlock = new BroadcastBlock<int>(i => i); // ⬅ Here
var processBlock1 = new TransformBlock<int, int>(x =>
DoSomethingAsync(5, "processBlock1")); //2
var processBlock2 = new TransformBlock<int, int>(x =>
DoSomethingAsync(2, "processBlock2")); //3
//var saveBlock =
// new ActionBlock<int>(
// x => Save(x)); //4
var saveBlockJoinedProcess =
new ActionBlock<Tuple<int, int>>(
x => SaveJoined(x.Item1, x.Item2)); //4
var saveBlockJoin = new JoinBlock<int, int>();
readBlock.LinkTo(braodcastBlock, new DataflowLinkOptions
{ PropagateCompletion = true });
braodcastBlock.LinkTo(processBlock1,
new DataflowLinkOptions { PropagateCompletion = true }); //5
braodcastBlock.LinkTo(processBlock2,
new DataflowLinkOptions { PropagateCompletion = true }); //6
processBlock1.LinkTo(
saveBlockJoin.Target1); //7
processBlock2.LinkTo(
saveBlockJoin.Target2); //8
saveBlockJoin.LinkTo(saveBlockJoinedProcess,
new DataflowLinkOptions { PropagateCompletion = true });
readBlock.Post(1); //10
//readBlock.Post(2); //10
Task.WhenAll(processBlock1.Completion,processBlock2.Completion)
.ContinueWith(_ => saveBlockJoin.Complete());
readBlock.Complete(); //12
saveBlockJoinedProcess.Completion.Wait(); //13
if (readBlock.Completion.Exception != null
|| saveBlockJoinedProcess.Completion.Exception != null
|| processBlock1.Completion.Exception != null
|| processBlock2.Completion.Exception != null)
{
throw readBlock.Completion.Exception;
}
}
private static int DoSomething(int i, string method)
{
Console.WriteLine($"Do Something, callng method : { method}");
throw new Exception("Fake Exception!");
return i;
}
private static async Task<int> DoSomethingAsync(int i, string method)
{
Console.WriteLine($"Do SomethingAsync");
throw new Exception("Fake Exception!");
await Task.Delay(new TimeSpan(0, 0, i));
Console.WriteLine($"Do Something : {i}, callng method : { method}");
return i;
}
private static void Save(int x)
{
Console.WriteLine("Save!");
}
private static void SaveJoined(int x, int y)
{
Thread.Sleep(new TimeSpan(0, 0, 10));
Console.WriteLine("Save Joined!");
}
}
}

I had a look online to see what's a suggested approach but didn't see anything obvious.
If you have a pipeline (more or less), then the common approach is to use PropagateCompletion to shut down the pipe. If you have more complex topologies, then you would need to complete blocks by hand.
In your case, you have an attempted propagation here:
Task.WhenAll(
processBlock1.Completion,
processBlock2.Completion)
.ContinueWith(_ => saveBlockJoin.Complete());
But this code will not propagate exceptions. When both processBlock1.Completion and processBlock2.Completion complete, saveBlockJoin is completed successfully.
A better solution would be to use await instead of ContinueWith:
async Task PropagateToSaveBlockJoin()
{
try
{
await Task.WhenAll(processBlock1.Completion, processBlock2.Completion);
saveBlockJoin.Complete();
}
catch (Exception ex)
{
((IDataflowBlock)saveBlockJoin).Fault(ex);
}
}
_ = PropagateToSaveBlockJoin();
Using await encourages you to handle exceptions, which you can do by passing them to Fault to propagate the exception.

Propagating errors backward in the pipeline is not supported in the TPL Dataflow out of the box, which is especially annoying when the blocks have a bounded capacity. In this case an error in a block downstream may cause the blocks in front of it to block indefinitely. The only solution I know is to use the cancellation feature, and cancel all blocks in case anyone fails. Here is how it can be done. First create a CancellationTokenSource:
var cts = new CancellationTokenSource();
Then create the blocks one by one, embedding the same CancellationToken in the options of all of them:
var options = new ExecutionDataflowBlockOptions()
{ BoundedCapacity = 10, CancellationToken = cts.Token };
var block1 = new TransformBlock<double, double>(Math.Sqrt, options);
var block2 = new ActionBlock<double>(Console.WriteLine, options);
Then link the blocks together, including the PropagateCompletion setting:
block1.LinkTo(block2, new DataflowLinkOptions { PropagateCompletion = true });
Finally use an extension method to trigger the cancellation of the CancellationTokenSource in case of an exception:
block1.OnFaultedCancel(cts);
block2.OnFaultedCancel(cts);
The OnFaultedCancel extension method is shown below:
public static class DataflowExtensions
{
public static void OnFaultedCancel(this IDataflowBlock dataflowBlock,
CancellationTokenSource cts)
{
dataflowBlock.Completion.ContinueWith(_ => cts.Cancel(), default,
TaskContinuationOptions.OnlyOnFaulted |
TaskContinuationOptions.ExecuteSynchronously, TaskScheduler.Default);
}
}

at the first look, if have only some minor points (not looking at your architecture). it seems to me that you have mixed some newer and some older constructs. and there are some code parts which are unnecessary.
for example:
private static void ProcessB()
{
Task.WhenAll(Task.Run(() => DoSomething(1, "ProcessB"))).Wait();
}
using the Wait()-method, if any exceptions happen, they will be wrapped in a System.AggregateException. in my opinion, this is better:
private static async Task ProcessBAsync()
{
await Task.Run(() => DoSomething(1, "ProcessB"));
}
using async-await, if an exception occurs, the await statement rethrows the first exception which is wrapped in the System.AggregateException. This allows you to try-catch for concrete exception types and handle only cases you really can handle.
another thing is this part of your code:
private static void ProcessA()
{
var random = new Random();
var readBlock = new TransformBlock<int, int>(
x =>
{
try { return DoSomething(x, "readBlock"); }
catch (Exception e)
{
throw e;
}
},
new ExecutionDataflowBlockOptions { MaxDegreeOfParallelism = 1 }); //1
Why catch an exception only to rethrow it? in this case, the try-catch is redundant.
And this here:
private static void SaveJoined(int x, int y)
{
Thread.Sleep(new TimeSpan(0, 0, 10));
Console.WriteLine("Save Joined!");
}
It is much better to use await Task.Delay(....). Using Task.Delay(...), your application will not freeze.

TaskFactory, Starting a new Task when one ends

I have found many methods of using the TaskFactory but I could not find anything about starting more tasks and watching when one ends and starting another one.
I always want to have 10 tasks working.
I want something like this
int nTotalTasks=10;
int nCurrentTask=0;
Task<bool>[] tasks=new Task<bool>[nThreadsNum];
for (int i=0; i<1000; i++)
{
string param1="test";
string param2="test";
if (nCurrentTask<10) // if there are less than 10 tasks then start another one
tasks[nCurrentThread++] = Task.Factory.StartNew<bool>(() =>
{
MyClass cls = new MyClass();
bool bRet = cls.Method1(param1, param2, i); // takes up to 2 minutes to finish
return bRet;
});
// How can I stop the for loop until a new task is finished and start a new one?
}

Check out the Task.WaitAny method:
Waits for any of the provided Task objects to complete execution.
Example from the documentation:
var t1 = Task.Factory.StartNew(() => DoOperation1());
var t2 = Task.Factory.StartNew(() => DoOperation2());
Task.WaitAny(t1, t2)

I would use a combination of Microsoft's Reactive Framework (NuGet "Rx-Main") and TPL for this. It becomes very simple.
Here's the code:
int nTotalTasks=10;
string param1="test";
string param2="test";
IDisposable subscription =
Observable
.Range(0, 1000)
.Select(i => Observable.FromAsync(() => Task.Factory.StartNew<bool>(() =>
{
MyClass cls = new MyClass();
bool bRet = cls.Method1(param1, param2, i); // takes up to 2 minutes to finish
return bRet;
})))
.Merge(nTotalTasks)
.ToArray()
.Subscribe((bool[] results) =>
{
/* Do something with the results. */
});
The key part here is the .Merge(nTotalTasks) which limits the number of concurrent tasks.
If you need to stop the processing part way thru just call subscription.Dispose() and everything gets cleaned up for you.
If you want to process each result as they are produced you can change the code from the .Merge(...) like this:
.Merge(nTotalTasks)
.Subscribe((bool result) =>
{
/* Do something with each result. */
});

This should be all you need, not complete, but all you need to do is wait on the first to complete and then run the second.
Task.WaitAny(task to wait on);
Task.Factory.StartNew()

Have you seen the BlockingCollection class? It allows you to have multiple threads running in parallel and you can wait from results from one task to execute another. See more information here.

The answer depends on whether the tasks to be scheduled are CPU or I/O bound.
For CPU-intensive work I would use Parallel.For() API setting the number of thread/tasks through MaxDegreeOfParallelism property of ParallelOptions
For I/O bound work the number of concurrently executing tasks can be significantly larger than the number of available CPUs, so the strategy is to rely on async methods as much as possible, which reduces the total number of threads waiting for completion.
How can I stop the for loop until a new task is finished and start a
new one?
The loop can be throttled by using await:
static void Main(string[] args)
{
var task = DoWorkAsync();
task.Wait();
// handle results
// task.Result;
Console.WriteLine("Done.");
}
async static Task<bool> DoWorkAsync()
{
const int NUMBER_OF_SLOTS = 10;
string param1="test";
string param2="test";
var results = new bool[NUMBER_OF_SLOTS];
AsyncWorkScheduler ws = new AsyncWorkScheduler(NUMBER_OF_SLOTS);
for (int i = 0; i < 1000; ++i)
{
await ws.ScheduleAsync((slotNumber) => DoWorkAsync(i, slotNumber, param1, param2, results));
}
ws.Complete();
await ws.Completion;
}
async static Task DoWorkAsync(int index, int slotNumber, string param1, string param2, bool[] results)
{
results[slotNumber] = results[slotNumber} && await Task.Factory.StartNew<bool>(() =>
{
MyClass cls = new MyClass();
bool bRet = cls.Method1(param1, param2, i); // takes up to 2 minutes to finish
return bRet;
}));
}
A helper class AsyncWorkScheduler uses TPL.DataFlow components as well as Task.WhenAll():
class AsyncWorkScheduler
{
public AsyncWorkScheduler(int numberOfSlots)
{
m_slots = new Task[numberOfSlots];
m_availableSlots = new BufferBlock<int>();
m_errors = new List<Exception>();
m_tcs = new TaskCompletionSource<bool>();
m_completionPending = 0;
// Initial state: all slots are available
for(int i = 0; i < m_slots.Length; ++i)
{
m_slots[i] = Task.FromResult(false);
m_availableSlots.Post(i);
}
}
public async Task ScheduleAsync(Func<int, Task> action)
{
if (Volatile.Read(ref m_completionPending) != 0)
{
throw new InvalidOperationException("Unable to schedule new items.");
}
// Acquire a slot
int slotNumber = await m_availableSlots.ReceiveAsync().ConfigureAwait(false);
// Schedule a new task for a given slot
var task = action(slotNumber);
// Store a continuation on the task to handle completion events
m_slots[slotNumber] = task.ContinueWith(t => HandleCompletedTask(t, slotNumber), TaskContinuationOptions.ExecuteSynchronously);
}
public async void Complete()
{
if (Interlocked.CompareExchange(ref m_completionPending, 1, 0) != 0)
{
return;
}
// Signal the queue's completion
m_availableSlots.Complete();
await Task.WhenAll(m_slots).ConfigureAwait(false);
// Set completion
if (m_errors.Count != 0)
{
m_tcs.TrySetException(m_errors);
}
else
{
m_tcs.TrySetResult(true);
}
}
public Task Completion
{
get
{
return m_tcs.Task;
}
}
void SetFailed(Exception error)
{
lock(m_errors)
{
m_errors.Add(error);
}
}
void HandleCompletedTask(Task task, int slotNumber)
{
if (task.IsFaulted || task.IsCanceled)
{
SetFailed(task.Exception);
return;
}
if (Volatile.Read(ref m_completionPending) == 1)
{
return;
}
// Release a slot
m_availableSlots.Post(slotNumber);
}
int m_completionPending;
List<Exception> m_errors;
BufferBlock<int> m_availableSlots;
TaskCompletionSource<bool> m_tcs;
Task[] m_slots;
}

Why TPL Dataflow block.LinkTo does not give any output?

I am quite new to the topic TPL Dataflow. In the book Concurrency in C# I tested the following example. I can't figure out why there's no output which should be 2*2-2=2;
static void Main(string[] args)
{
//Task tt = test();
Task tt = test1();
Console.ReadLine();
}
static async Task test1()
{
try
{
var multiplyBlock = new TransformBlock<int, int>(item =>
{
if (item == 1)
throw new InvalidOperationException("Blech.");
return item * 2;
});
var subtractBlock = new TransformBlock<int, int>(item => item - 2);
multiplyBlock.LinkTo(subtractBlock,
new DataflowLinkOptions { PropagateCompletion = true });
multiplyBlock.Post(2);
await subtractBlock.Completion;
int temp = subtractBlock.Receive();
Console.WriteLine(temp);
}
catch (AggregateException e)
{
// The exception is caught here.
foreach (var v in e.InnerExceptions)
{
Console.WriteLine(v.Message);
}
}
}
Update1: I tried another example. Still I did not use Block.Complete() but I thought when the first block's completed, the result is passed into the second block automatically.
private static async Task test3()
{
TransformManyBlock<int, int> tmb = new TransformManyBlock<int, int>((i) => { return new int[] {i, i + 1}; });
ActionBlock<int> ab = new ActionBlock<int>((i) => Console.WriteLine(i));
tmb.LinkTo(ab);
for (int i = 0; i < 4; i++)
{
tmb.Post(i);
}
//tmb.Complete();
await ab.Completion;
Console.WriteLine("Finished post");
}

This part of the code:
await subtractBlock.Completion;
int temp = subtractBlock.Receive();
is first (asynchronously) waiting for the subtraction block to complete, and then attempting to retrieve an output from the block.
There are two problems: the source block is never completed, and the code is attempting to retrieve output from a completed block. Once a block has completed, it will not produce any more data.
(I assume you're referring to the example in recipe 4.2, which will post 1, causing the exception, which completes the block in a faulted state).
So, you can fix this test by completing the source block (and the completion will propagate along the link to the subtractBlock automatically), and by reading the output before (asynchronously) waiting for subtractBlock to complete:
multiplyBlock.Complete();
int temp = subtractBlock.Receive();
await subtractBlock.Completion;

C# infinitive task loop using Task<> class + cancellation

I`m trying to make a small class for the multithreading usage in my WinForm projects.
Tried Threads(problems with UI), Backgroundworker(smth went wrong with UI too, just leave it now:)), now trying to do it with Task class. But now, can`t understand, how to make an infinitive loop and a cancelling method (in class) for all running tasks.
Examples i found is to be used in 1 method.
So, here is a structure & code of currently working part (Worker.css and methonds used in WinForm code).
Worker.css
class Worker
{
public static int threadCount { get; set; }
public void doWork(ParameterizedThreadStart method)
{
Task[] tasks = Enumerable.Range(0, 4).Select(i => Task.Factory.StartNew(() => method(i))).ToArray();
}
}
usage on
Form1.cs
private void Start_btn_Click(object sender, EventArgs e)
{
Worker.threadCount = 1; //actually it doesn`t using now, number of tasks is declared in class temporaly
Worker worker = new Worker();
worker.doWork(Job);
string logString_1 = string.Format("Starting {0} threads...", Worker.threadCount);
log(logString_1);
}
public static int j = 0;
private void Job(object sender)
{
Worker worker = new Worker();
Random r = new Random();
log("Thread "+Thread.CurrentThread.ManagedThreadId +" is working...");
for (int i = 0; i < 5; i++)
{
j++;
log("J==" + j);
if (j == 50)
{
//worker.Stop();
log("STOP");
}
}
Thread.Sleep(r.Next(500, 1000));
}
So, it run an example 4 threads, they executed, i got J==20 in my log, it`s ok.
My question is, how to implement infinitive loop for the tasks, created by Worker.doWork() method.
And also to make a .Stop() method for the Worker class (which should just stop all tasks when called). As i understand it`s related questions, so i put it in 1.
I tryed some solutions, but all of them based on the CancellationToken usage, but i have to create this element only inside of the Worker.doWork() method, so i can`t use the same token to create a Worker.Stop() method.
Someone can help? threads amount range i have to use in this software is about 5-200 threads.
using J computation is just an example of the the easy condition used to stop a software work(stop of tasks/threads).
In real, stop conditions is mostly like Queue<> is finished, or List<> elements is empty(finished).

Finally, get it works.
class Worker
{
public static int threadCount { get; set; }
Task[] tasks;
//ex data
public static string exception;
static CancellationTokenSource wtoken = new CancellationTokenSource();
CancellationToken cancellationToken = wtoken.Token;
public void doWork(ParameterizedThreadStart method)
{
try
{
tasks = Enumerable.Range(0, 4).Select(i => Task.Factory.StartNew(() =>
{
while (!cancellationToken.IsCancellationRequested)
{
method(i);
}
}, cancellationToken)).ToArray();
}
catch (Exception ex) { exception = ex.Message; }
}
public void HardStop()
{
try
{
using (wtoken)
{
wtoken.Cancel();
}
wtoken = null;
tasks = null;
}
catch (Exception ex) { exception = ex.Message; }
}
}
But if i`m using this method to quit cancellationToken.ThrowIfCancellationRequested();
Get a error:
when Job() method reach J == 50, and worker.HardStop() function called, program window crashes and i get and exception "OparetionCanceledException was unhandled by user code"
on this string
cancellationToken.ThrowIfCancellationRequested();
so, whats wrong? i`m already put it in try{} catch(){}
as i understood, just some boolean properties should be changed in Task (Task.IsCancelled == false, Task.IsFaulted == true) on wtoken.Cancel();

I'd avoid all of the mucking around with tasks and use Microsoft's Reactive Framework (NuGet "Rx-Main") for this.
Here's how:
var r = new Random();
var query =
Observable
.Range(0, 4, Scheduler.Default)
.Select(i =>
Observable
.Generate(0, x => true, x => x, x => x,
x => TimeSpan.FromMilliseconds(r.Next(500, 1000)),
Scheduler.Default)
.Select(x => i))
.Merge();
var subscription =
query
.Subscribe(i => method(i));
And when you want to cancel the calls to method just do this:
subscription.Dispose();
I've tested this and it works like a treat.
If I wrap this up in your worker class then it looks like this:
class Worker
{
private Random _r = new Random();
private IDisposable _subscription = null;
public void doWork()
{
_subscription =
Observable
.Range(0, 4, Scheduler.Default)
.Select(n =>
Observable
.Generate(
0, x => true, x => x, x => x,
x => TimeSpan.FromMilliseconds(_r.Next(500, 1000)),
Scheduler.Default)
.Select(x => n))
.Merge()
.Subscribe(i => method(i));
}
public void HardStop()
{
_subscription.Dispose();
}
}

Is it possible always to force a new thread with Task?

I am trying to create a new thread each time Task.Factory.StartNew is called. The question is how to run the code bellow without throwing the exception:
static void Main(string[] args)
{
int firstThreadId = 0;
Task.Factory.StartNew(() => firstThreadId = Thread.CurrentThread.ManagedThreadId);
for (int i = 0; i < 100; i++)
{
Task.Factory.StartNew(() =>
{
while (true)
{
Thread.Sleep(1000);
if (firstThreadId == Thread.CurrentThread.ManagedThreadId)
throw new Exception("The first thread is reused.");
}
});
}
Console.Read();
}
EDIT: the new code if you comment the first for statement there is no problem. But if you have it, WOW, the message "Thread reused" is written to the console. Can you explain that because I am really confused.
static void Main(string[] args)
{
ConcurrentDictionary<int, int> startedThreads = new ConcurrentDictionary<int, int>();
for (int i = 0; i < 10; i++)
{
Task.Factory.StartNew(() =>
{
Task.Factory.StartNew(() =>
{
startedThreads.AddOrUpdate(Thread.CurrentThread.ManagedThreadId,
Thread.CurrentThread.ManagedThreadId, (a, b) => b);
}, TaskCreationOptions.LongRunning);
for (int j = 0; j < 100; j++)
{
Task.Factory.StartNew(() =>
{
while (true)
{
Thread.Sleep(10);
if (startedThreads.ContainsKey(
Thread.CurrentThread.ManagedThreadId))
Console.WriteLine("Thread reused");
}
}, TaskCreationOptions.LongRunning);
}
});
}
Console.Read();
}

If you specify TaskCreationOptions.LongRunning when starting the task, that provides a hint to the scheduler, which the default scheduler takes as an indicator to create a new thread for the task.
It's only a hint - I'm not sure I'd rely on it... but I haven't seen any counterexamples using the default scheduler.

Adding to Jon Skeet's answer, if you want to guarantee that a new thread is created every time, you can write your own TaskScheduler that creates a new thread.

Try this:
var taskCompletionSource = new TaskCompletionSource<bool>();
Thread t = new Thread(() =>
{
try
{
Operation();
taskCompletionSource.TrySetResult(true);
}
catch (Exception e)
{
taskCompletionSource.TrySetException(e);
}
});
void Operation()
{
// Some work in thread
}
t.Start();
await taskCompletionSource.Task;
You also can write extension methods for Action, Func and so on.
For example:
public static Task RunInThread(
this Action action,
Action<Thread> initThreadAction = null)
{
TaskCompletionSource<bool> taskCompletionSource = new TaskCompletionSource<bool>();
Thread thread = new Thread(() =>
{
try
{
action();
taskCompletionSource.TrySetResult(true);
}
catch (Exception e)
{
taskCompletionSource.TrySetException(e);
}
});
initThreadAction?.Invoke(thread);
thread.Start();
return taskCompletionSource.Task;
}
or
public static Task<TResult> RunInThread<T1, T2, TResult>(
this Func<T1, T2, TResult> function,
T1 param1,
T2 param2,
Action<Thread> initThreadAction = null)
{
TaskCompletionSource<TResult> taskCompletionSource = new TaskCompletionSource<TResult>();
Thread thread = new Thread(() =>
{
try
{
TResult result = function(param1, param2);
taskCompletionSource.TrySetResult(result);
}
catch (Exception e)
{
taskCompletionSource.TrySetException(e);
}
});
initThreadAction?.Invoke(thread);
thread.Start();
return taskCompletionSource.Task;
}
and use it like that:
var result = await some_function.RunInThread(param1, param2).ConfigureAwait(true);

Hello and thank you all for the answers. You all got +1. All suggested solution did not work for my case. The problem is that when you sleep a thread it will be reused at some point of time. The people above suggested:
using LongRunning => This will not work if you have nested/child
tasks
custom task scheduler => I tried to write my own and also tried this
ThreadPerTaskScheduler which also di not work.
using pure threads => Still failing...
you could also check this project at Multithreading.Scheduler github
My solution
I don't like it but it works. Basically I block the thread so it cannot be reused. Bellow are the extension methods and a working example. Again, thank you.
https://gist.github.com/4150635
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
namespace ConsoleApplication
{
public static class ThreadExtensions
{
/// <summary>
/// Blocks the current thread for a period of time so that the thread cannot be reused by the threadpool.
/// </summary>
public static void Block(this Thread thread, int millisecondsTimeout)
{
new WakeSleepClass(millisecondsTimeout).SleepThread();
}
/// <summary>
/// Blocks the current thread so that the thread cannot be reused by the threadpool.
/// </summary>
public static void Block(this Thread thread)
{
new WakeSleepClass().SleepThread();
}
/// <summary>
/// Blocks the current thread for a period of time so that the thread cannot be reused by the threadpool.
/// </summary>
public static void Block(this Thread thread, TimeSpan timeout)
{
new WakeSleepClass(timeout).SleepThread();
}
class WakeSleepClass
{
bool locked = true;
readonly TimerDisposer timerDisposer = new TimerDisposer();
public WakeSleepClass(int sleepTime)
{
var timer = new Timer(WakeThread, timerDisposer, sleepTime, sleepTime);
timerDisposer.InternalTimer = timer;
}
public WakeSleepClass(TimeSpan sleepTime)
{
var timer = new Timer(WakeThread, timerDisposer, sleepTime, sleepTime);
timerDisposer.InternalTimer = timer;
}
public WakeSleepClass()
{
var timer = new Timer(WakeThread, timerDisposer, Timeout.Infinite, Timeout.Infinite);
timerDisposer.InternalTimer = timer;
}
public void SleepThread()
{
while (locked)
lock (timerDisposer) Monitor.Wait(timerDisposer);
locked = true;
}
public void WakeThread(object key)
{
locked = false;
lock (key) Monitor.Pulse(key);
((TimerDisposer)key).InternalTimer.Dispose();
}
class TimerDisposer
{
public Timer InternalTimer { get; set; }
}
}
}
class Program
{
private static readonly Queue<CancellationTokenSource> tokenSourceQueue = new Queue<CancellationTokenSource>();
static void Main(string[] args)
{
CancellationTokenSource tokenSource = new CancellationTokenSource();
tokenSourceQueue.Enqueue(tokenSource);
ConcurrentDictionary<int, int> startedThreads = new ConcurrentDictionary<int, int>();
for (int i = 0; i < 10; i++)
{
Thread.Sleep(1000);
Task.Factory.StartNew(() =>
{
startedThreads.AddOrUpdate(Thread.CurrentThread.ManagedThreadId, Thread.CurrentThread.ManagedThreadId, (a, b) => b);
for (int j = 0; j < 50; j++)
Task.Factory.StartNew(() => startedThreads.AddOrUpdate(Thread.CurrentThread.ManagedThreadId, Thread.CurrentThread.ManagedThreadId, (a, b) => b));
for (int j = 0; j < 50; j++)
{
Task.Factory.StartNew(() =>
{
while (!tokenSource.Token.IsCancellationRequested)
{
if (startedThreads.ContainsKey(Thread.CurrentThread.ManagedThreadId)) Console.WriteLine("Thread reused");
Thread.CurrentThread.Block(10);
if (startedThreads.ContainsKey(Thread.CurrentThread.ManagedThreadId)) Console.WriteLine("Thread reused");
}
}, tokenSource.Token, TaskCreationOptions.LongRunning, TaskScheduler.Default)
.ContinueWith(task =>
{
WriteExceptions(task.Exception);
Console.WriteLine("-----------------------------");
}, TaskContinuationOptions.OnlyOnFaulted);
}
Thread.CurrentThread.Block();
}, tokenSource.Token, TaskCreationOptions.LongRunning, TaskScheduler.Default)
.ContinueWith(task =>
{
WriteExceptions(task.Exception);
Console.WriteLine("-----------------------------");
}, TaskContinuationOptions.OnlyOnFaulted);
}
Console.Read();
}
private static void WriteExceptions(Exception ex)
{
Console.WriteLine(ex.Message);
if (ex.InnerException != null)
WriteExceptions(ex.InnerException);
}
}
}

Just start threads with new Thread() and then Start() them
static void Main(string[] args)
{
ConcurrentDictionary<int, int> startedThreads = new ConcurrentDictionary<int, int>();
for (int i = 0; i < 10; i++)
{
new Thread(() =>
{
new Thread(() =>
{
startedThreads.AddOrUpdate(Thread.CurrentThread.ManagedThreadId, Thread.CurrentThread.ManagedThreadId, (a, b) => b);
}).Start();
for (int j = 0; j < 100; j++)
{
new Thread(() =>
{
while (true)
{
Thread.Sleep(10);
if (startedThreads.ContainsKey(Thread.CurrentThread.ManagedThreadId))
Console.WriteLine("Thread reused");
}
}).Start();
}
}).Start();
}
Console.Read();
}
Tasks are supposed to be managed by the scheduler. The whole idea of Tasks is that the runtime will decide when a new thread is needed. On the other hand if you do need different threads chances are something else in the code is wrong like overdependency on Thread.Sleep() or thread local storage.
As pointed out you can create your own TaskScheduler and use tasks to create threads but then why use Tasks to begin with?

Here is a custom TaskScheduler that executes the tasks on a dedicated thread per task:
public class ThreadPerTask_TaskScheduler : TaskScheduler
{
protected override void QueueTask(Task task)
{
var thread = new Thread(() => TryExecuteTask(task));
thread.IsBackground = true;
thread.Start();
}
protected override bool TryExecuteTaskInline(Task task, bool taskWasPreviouslyQueued)
{
return TryExecuteTask(task);
}
protected override IEnumerable<Task> GetScheduledTasks() { yield break; }
}
Usage example:
var parallelOptions = new ParallelOptions()
{
MaxDegreeOfParallelism = 3,
TaskScheduler = new ThreadPerTask_TaskScheduler()
};
Parallel.ForEach(Enumerable.Range(1, 10), parallelOptions, item =>
{
Console.WriteLine($"{DateTime.Now:HH:mm:ss.fff}" +
$" [{Thread.CurrentThread.ManagedThreadId}]" +
$" Processing #{item}" +
(Thread.CurrentThread.IsBackground ? ", Background" : "") +
(Thread.CurrentThread.IsThreadPoolThread ? ", ThreadPool" : ""));
Thread.Sleep(1000); // Simulate CPU-bound work
});
Output:
20:38:56.770 [4] Processing #3, Background
20:38:56.770 [5] Processing #2, Background
20:38:56.770 [1] Processing #1
20:38:57.782 [1] Processing #4
20:38:57.783 [8] Processing #5, Background
20:38:57.783 [7] Processing #6, Background
20:38:58.783 [1] Processing #7
20:38:58.783 [10] Processing #8, Background
20:38:58.787 [9] Processing #9, Background
20:38:59.783 [1] Processing #10
Try it on Fiddle.
This custom TaskScheduler allows the current thread to participate in the computations too. This is demonstrated in the above example by the thread [1] processing the items #1, #4, #7 and #10. If you don't want this to happen, just replace the code inside the TryExecuteTaskInline with return false;.
Another example, featuring the Task.Factory.StartNew method. Starting 100 tasks on 100 different threads:
var oneThreadPerTask = new ThreadPerTask_TaskScheduler();
Task[] tasks = Enumerable.Range(1, 100).Select(_ =>
{
return Task.Factory.StartNew(() =>
{
Thread.Sleep(1000); // Simulate long-running work
}, default, TaskCreationOptions.None, oneThreadPerTask);
}).ToArray();
In this case the current thread is not participating in the work, because all tasks are started behind the scenes by invoking their Start method, and not the
RunSynchronously.