Depending on what you are asking, you may not need the maximum queue size. Starting with version 2.0, RabbitMQ will easily save large queues on disk, rather than storing all messages in RAM. Therefore, if you are worried that the broker is crashing because it is running out of resources, this is actually not a big problem in most cases - if you are not tied to a place on your hard drive.
On the whole, this perseverance actually has very little effect on performance, because by definition the only "hot" parts of the line are the head and tail, which remain in RAM; most of the backlog is βcold,β so itβs not much different from sitting on the disk instead.
Recently, we found that with high throughput, this is not so simple - in some circumstances, throughput may deteriorate as the queue grows, which can lead to unlimited growth in the queue. But when this happens, this is a processor function, and for a long time we walked on it without a blow.
source share