I have a service and it can process only one request at a time. I deployed service with kubernetes.

I want this: Every request will be put in a RabbitMQ queue. When the top request in the queue is sent, a new pod will create. The pod will be destroyed after the service at docker container ends. A new pod will open for the new request. Also, there will be a limit for pod count. New request in queue will be held when the limit is reached.

Can you guide me for this use case???