Tutorial six: consumers are accumulated between recoveries

[Crabzmatic]

The field _recordedConsumers in AutorecoveryConnection keeps getting more populated. After couple of thousands messages there is a large amount of accumulated consumers. With large amount of messages this can grow to a noticable number (hundreds of MB).

Edit: https://www.rabbitmq.com/tutorials/tutorial-six-dotnet.html this tutorial is also a reproducer. Sending lots of requests like this:

while (true)
{
    Console.WriteLine(" [x] Requesting fib(30)");
    var response = rpcClient.Call("30");
    Console.WriteLine(" [.] Got '{0}'", response);

    await Task.Delay(20);
}

accumulates 10 000 RecordedConsumers in under 10 minutes. These RecordedConsumers appear to linger there forever for long running connections. This can cause problems for connections that run for a long time.

[bollhals]

That tutorial just seems wrong to me. Why does it set up everything in the constructor except the BasicPublish (which needs to be in Call) and the BasicConsume call. If you move it to the constructor it should be fine.

BasicConsume starts consuming the messages in the queue by the consumer. The AutorecoveryConnection remembers which one was started (to start consuming again after a reconnection). If these consumers are never cancelled, they grow endlessly. (Which they do in the tutorial code)

[michaelklishin]

@bollhals thank you for taking a look. Executable versions of the tutorials are open source, @Crabzmatic or anyone else is welcome to submit a PR :) The website version is in the website repo which is also public.