Using a System.Threading.Timer
results in threads being spun from a ThreadPool
, which means if the interval of execution for the timer expires while a thread is still processing by order of a previous request, then the same callback will be delegated to execute on another thread. This is obviously going to cause problems in most cases unless the callback is re-entrant aware, but I'm wondering how to go about it the best (meaning safe) way.
Let's say we have the following:
ReaderWriterLockSlim OneAtATimeLocker = new ReaderWriterLockSlim();
OneAtATimeCallback = new TimerCallback(OnOneAtATimeTimerElapsed);
OneAtATimeTimer = new Timer(OneAtATimeCallback , null, 0, 1000);
Should the whole shebang be be locked down, as such:
private void OnOneAtATimeTimerElapsed(object state)
{
if (OneAtATimeLocker.TryEnterWriteLock(0))
{
//get real busy for two seconds or more
OneAtATimeLocker.ExitWriteLock();
}
}
Or, should only entry be managed, and kick out 'trespassers', as such:
private void OnOneAtATimeTimerElapsed(object state)
{
if (!RestrictOneAtATime())
{
return;
}
//get real busy for two seconds or more
if(!ReleaseOneAtATime())
{
//Well, Hell's bells and buckets of blood!
}
}
bool OneAtATimeInProgress = false;
private bool RestrictToOneAtATime()
{
bool result = false;
if (OneAtATimeLocker.TryEnterWriteLock(0))
{
if(!OneAtATimeInProgress)
{
OneAtATimeInProgress = true;
result = true;
}
OneAtATimeLocker.ExitWriteLock();
}
return result;
}
private bool ReleaseOneAtATime()
{
bool result = false;
//there shouldn't be any 'trying' about it...
if (OneAtATimeLocker.TryEnterWriteLock(0))
{
if(OneAtATimeInProgress)
{
OneAtATimeInProgress = false;
result = true;
}
OneAtATimeLocker.ExitWriteLock();
}
return result;
}
Does the first have any performance implications because it locks for the extent of the method?
Does the second even offer the safety one might think it does - is something escaping me?
Are there any other ways to go about this reliably, and preferably?
Timer is not thread-safe. Since then this has been repeated on blogs, in Richter's book "CLR via C#", on SO, but this is never justified.
The Timer component is a server-based timer that raises an Elapsed event in your application after the number of milliseconds in the Interval property has elapsed. You can configure the Timer object to raise the event just once or repeatedly using the AutoReset property.
In order to make a thread as thread safe there are some thread synchronization techniques like Monitor/Lock, Mutex, Semaphore and SemaphoreSlim using these techniques we can achieve Thread Safety.
start() is a function that is used to initialize a timer. To end or quit the timer, one must use a cancel() function. Importing the threading class is necessary for one to use the threading class. The calling thread can be suspended for seconds using the function time.
Lots of ways to deal with this. A simple way is to just not make the timer periodic, make it a one shot by only setting the dueTime argument. Then re-enable the timer in the callback in a finally block. That guarantees that the callback cannot run concurrently.
This is of course makes the interval variable by the execution time of the callback. If that's not desirable and the callback only occasionally takes longer than the timer period then a simple lock will get the job done. Yet another strategy is Monitor.TryEnter and just give up on the callback if it returns false. None of these are particularly superior, pick what you like best.
A slightly different riff on the same concept, if you do not have to process every tick. This simply bails on the current timer tick if you are still processing a previous tick.
private int reentrancyCount = 0;
...
OnTick()
{
try
{
if (Interlocked.Increment(ref reentrancyCount) > 1)
return;
// Do stuff
}
finally
{
Interlocked.Decrement(ref reentrancyCount);
}
}
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With