Not sure what you are refering to when you say that it doesnt consume the queue.
I use RabbitMQ a lot in workflows and it working perfectly with the nodes provided.
There is 2 nodes, one normal and one trigger. You of course would need the trigger node to start consuming the queue.
the node consumes the queue but only from the moment it is 100% active, if the node is unavailable or is unable to process all messages, those that are stopped in the queue will not be processed by the node as it is reactive
@Jon He said in another topic that this would be a future feature, while we don’t think about doing something using code
Didn’t get the point. The trigger node starts to consume the queue as early as you will activate the workflow. The node can’t be unavailable, only the broker could, but in this way, you won’t get any messages from it either. The node is able to process everything you would ask it to process. It depends on how you are going to design the error handling (flush the message on error / keep it unless the workflow ends and so on).
The topic you have mention is about the delaying the message processing