Rabbit Mq Publisher Nack behavior

[Pchase]

In this tutorial

https://www.rabbitmq.com/tutorials/tutorial-seven-dotnet.html

in section:

Strategy #3: Handling Publisher Confirms Asynchronously

It has a nack for one item but cleans all items below the deliveryTag.  Is this the correct behavior?  I will only get a nack for one item but all items below it are essentially ack'd?

channel.BasicNacks += (sender, ea) => { 

   outstandingConfirms.TryGetValue(ea.DeliveryTag, out string body);

   cleanOutstandingConfirms(ea.DeliveryTag, ea.Multiple); 

};  

[Luke Bakken]

Hello,

The ea instance has a Multiple field that indicates if the Nack is for every delivery tag up to the one in the object, or just the one.

https://github.com/rabbitmq/rabbitmq-tutorials/blob/master/dotnet/PublisherConfirms/PublisherConfirms.cs#L80-L108

Note that the value of Multiple is taken into account.

Thanks,

Luke

[Pchase]

The example shows it only tries to do something with the current deliveryTag.  Shouldn't the example show that it tries to do something with all the items below it instead of clearing them out?

var nacked = outstandingConfirms.Where(k => k.Key <= sequenceNumber);

foreach (var entry in nacked)
{

   outstandingConfirms.TryRemove(entry.Key, out x); 

   Console.WriteLine($"Message has been nack-ed. Sequence number: {entry.Key}, multiple: {ea.Multiple}");  
     

[Luke Bakken]

Hello,

Please remember that this is example code. We have to strike a balance between understand-ability and completeness.

A real application would possibly re-try messages, raise exceptions, etc. We can't cover all use cases. Real applications would keep a sliding window of outstanding confirmations and throttle publishers if the window is full. Each outstanding confirm could have a deadline after which a decision should be made as to re-publishing, discarding or otherwise dealing with the missed confirmation.

In general if your application starts missing confirmations something is awry with your environment.

Thanks,

Luke