Stop threads gracefuly

Every time you are using threads in your code you’ll need a way to stop them and from what I have been experienced the Thread.Abort() is not a choice. Thread.Abort will allways throw ThreadAbortException and what is worse is that the throw is asynchronous so a simple try/catch will not be helpful. On the Abort method the MSDN tells us that “the thread is not guaranteed to abort immediately, or at all” and better yet “If two calls to Abort come at the same time, it is possible that one call set the state information and the other call executes the Abort. However, an application cannot distinguish this.”.

In order to stop a worker in a graceful way you’ll need to use a volatile member in the worker class. My implementation follows:

ReadyForThreading class:

public abstract class ReadyForThreading
{
private readonly object locker = new object();
private bool stopRequired = false;
private bool isStopped = false;

public bool IsStopped
{
get
{
lock (locker) { return isStopped; }
}
protected set
{
lock (locker) { isStopped = value; }
}
}

public bool IsStopping
{
get
{
lock (locker) { return stopRequired; }
}
}

public void Stop()
{
lock (locker) { this.stopRequired = true; }
}
}

Now all you need to do is design your worker class based on the ReadyForThreading:

Worker example class:

public class SomeWorker : ReadyForThreading
{
public void DoWork(object state)
{
try
{
int i = 0;
while (!base.IsStopping)
{
// do work
Thread.Sleep(2000);
Console.WriteLine(“Working on {0} part 1 loop {1}”,
state.ToString(), i);

if (base.IsStopping)
{
break;
}

//do more work
Thread.Sleep(3000);
Console.WriteLine(“Work part 2 loop ” + i);

i++;
}
}
finally
{
base.IsStopped = true;
}
}
}

Usage example:

SomeWorker worker = new SomeWorker();
Thread th = new Thread(new ParameterizedThreadStart(worker.DoWork));
th.Start(“SomeObject”);
Console.WriteLine(“Press any key to stop the worker”);
Console.ReadLine();
worker.Stop();

---

Possibly related posts: (automatically generated)

• while(currentThread == backgroundThread)
• Multi-Threading Kernel for Data Integration