erlang - How to ensure RabbitMQ comes back on line when all cluster nodes are shutdown - Stack Overflow

admin管理员组

文章数量:1344403

I am new to RabbitMQ and am trying to implement a clustered node setup. We are using RabbitMQ 3.11.11 with Erlang 25.3. I also have an HA policy to mirror queues in the cluster with the following settings:

Basically we have an architecture running a multinode client/server model where each server is also running RabbitMQ (along with other services) and each server can be self-standing. One server is designated as the "primary" where other servers are "auxiliary" and can take over primary status for a variety of reasons.

With this architecture, servers can be brought down and come up in various orders and from my reading, RabbitMQ really wants the last node in the cluster that goes down to be the first that comes back up. I can't guarantee that happens.

Is there a way to tell RabbitMQ that when all nodes are down and one comes up (whether it was the last node or not) to just let it come up and then allow any other node to come up and rejoin the cluster? I have read about the force_boot command but I am not really sure when and how to use it (does the service need to be running? Does the app need to be running? Should I use it when all nodes come up? Go down? Etc) or is there another way to do what I want to do?

Lastly, I realize this might cause messages to be lost.

Thank you.

本文标签： erlangHow to ensure RabbitMQ comes back on line when all cluster nodes are shutdownStack Overflow