Consume is repeated indefinetely when raised OperationCancelledException

Is this a bug report?

Yes

Can you also reproduce the problem with the latest version?

Yes

Environment

1. Operating system: Windows 10 Enterprise
2. Visual Studio version: 2017, v15.9.8
3. Dotnet version: NetCore SDK 2.2.104

Steps to Reproduce

1. Bus configured with RabbitMQ transport
2. Retry policy on ‘ReceiveEndpoint’ set to Immediate(1)
3. Publish one event
4. Consumer throws TaskCancellationException or OperationCancelledException

Expected Behavior

The system calls consumer trying to consume message, then retry policy reiterate it exactly one time (UseMessageRetry(r => r.Immediate(1)))

Actual Behavior

The system calls consumer trying to consume message without end. The same scenario with ‘InMemory’ transport works as we expected.

Reproducible Demo

    public class Program
    {
        static void Main(string[] args)
        {
            var bus = Bus.Factory.CreateUsingRabbitMq(cfg =>
            {
                cfg.Host(new Uri("rabbitmq://localhost/"), h =>
                {
                    h.Username("guest");
                    h.Password("guest");
                });

                cfg.ReceiveEndpoint("input-queue", e =>
                {
                    e.UseMessageRetry(r => r.Immediate(1));

                    e.Consumer<MyConsumer>();
                });
            });
            bus.Start();

            bus.Publish(new SimpleEvent { Id = 123 });

            Console.ReadLine();
        }
    }
    public class MyConsumer : IConsumer<SimpleEvent>
    {
        public Task Consume(ConsumeContext<SimpleEvent> context)
        {
            var attempt = context.GetRetryAttempt();
            throw new TaskCanceledException("task cancelled");
        }
    }
    public class SimpleEvent
    {
        public int Id { get; set; }
    }

[newpma]

Thank you for quick response. What do you recommend to avoid infinite loop in case, that we actually faced, when consumer calls http api that is not accessible due to some network issue. I know, easy answer is to solve network issue, but still it could sometimes happen. What I missed in above ‘Reproducible Demo’ is that our abstraction over http client wraps exceptions.

    public class Program
    {
        static void Main(string[] args)
        {
            var bus = Bus.Factory.CreateUsingRabbitMq(cfg =>
            {
                cfg.Host(new Uri("rabbitmq://localhost/"), h =>
                {
                    h.Username("guest");
                    h.Password("guest");
                });

                cfg.ReceiveEndpoint("input-queue", e =>
                {
                    e.UseMessageRetry(r => 
                    {
                        r.Immediate(1);
                    });

                    e.Consumer<MyConsumer>();
                });
            });
            bus.Start();

            bus.Publish(new SimpleEvent { Id = 123 });

            Console.ReadLine();
        }
    }
    public class MyConsumer : IConsumer<SimpleEvent>
    {
        public Task Consume(ConsumeContext<SimpleEvent> context)
        {
            var attempt = context.GetRetryAttempt();
            throw new MyException("my exception", new TaskCanceledException("task cancelled"));
        }
    }
    public class SimpleEvent
    {
        public int Id { get; set; }
    }
    public class MyException : ApplicationException
    {
        public MyException(string message, Exception inner)
            : base(message, inner)
        {
        }
    }

[phatboyg]

This was addressed in v6.