C# - 超时后终止任务

Question

I have this method ExecuteAllAsync() which uses the field List jobs and executes asynchronously all Jobs in the list. I want to set a timeout to kill the job and stop the execution if it takes to long or no longer necessary. I tried doing so in the following code:

        //Execute all jobs async
        public async void ExecuteAllAsync()
        {
            if (jobs.Count()==0)
                return;
            List<Task> tasks = new List<Task>();
            foreach (Job job in jobs.ToArray())
            {
                int timeout = 500; 

                Task t = Task.Run(() => { 
                     ExecuteJob(job) // Do Work Here
                });
                tasks.Add(t);
                if (!t.Wait(timeout))
                     //Log Timeout Error
            }
            await Task.WhenAll(tasks);
        }

This doesn't work. I tried to also put Thread.Sleep(1000) in the ExecuteJob() Function to make all tasks timeout and and the result is that only some of the jobs timeout and some execute fully. I also use stopwatch to check how long the job takes and I see some of them takes more than 1 second.

Currently the timeout is set 500 ms but in the future each task will have different timeout value:

int timeout = (int)(job.TimeToKill() - DateTime.Now).TotalMilliseconds

How do I execute all jobs in list asynchronously while setting a timeout to each task? Thanks in advance.

Answer 1

You could use a CancellationToken :

You can create one with a CancellationTokenSource and then pass a timeout to .CancelAfter(int) ;

https://learn.microsoft.com/en-us/do.net/api/system.threading.cancellationtokensource.cancelafter?view.netcore-3.1

Make sure to check the cancellationToken inside your task.

Answer 2

You can use this simple approach:

public void ExecuteAll() //Note: since ExecuteJob is not async, this method not needs to be async anymore
{
    if (jobs.Count() == 0)
        return;

    int timeout = 500; 

    Parallel.ForEach(jobs, job => ExecuteJob(job, timeout);
}


public void ExecuteJob(Job job, int tiemout) //Your new ExecuteJob overload receiving a Timeout and using a CancellationToken internally
{
    using (var cts = new CancellationTokenSource(timeout))
    {
        foreach (var jobItem in job.ItemsToProcess) //Assuming you have something to do inside our job
        {
            if (cts.Token.IsCancellationRequested)
            {
                // ==> Log Timeout Error; cancel your job 
            }

            // Do your job
        }
    }
}

Note you can simply use Parallel.ForEach , it would take care for you about the multiple threads, optimizing them according to the numbers of cores available in the executing machine