RabbitMQ brokers should use cluster deployment mode

How to fix it

• Replaces the resource: A broker's deployment mode is fixed at creation: create a cluster broker with this one's users, queues and policies, repoint clients, drain or shovel the messages, then delete it.
• Can break clients: Cluster nodes restart one at a time during maintenance and drop their connections, so clients that do not reconnect automatically lose the broker.

In the console

Amazon MQ → Brokers → Create brokers → RabbitMQ → Select deployment mode (fixed at creation)

Deployment mode

Cluster deployment

Broker instance type

An mq.m7g type, mq.m7g.large or larger for production

RabbitMQ access

An administrator user; then recreate the flagged broker's users, virtual hosts, queues and policies

With the AWS CLI

aws mq create-broker --broker-name <new-broker-name> --engine-type RABBITMQ --engine-version <engine-version> --deployment-mode CLUSTER_MULTI_AZ --host-instance-type <instance-type> --auto-minor-version-upgrade --no-publicly-accessible --subnet-ids <subnet-id> --users Username=<username>,Password=<password> --region <region>

With Terraform

resource "aws_mq_broker" "cluster" {
  broker_name                = var.broker_name
  engine_type                = "RabbitMQ"
  engine_version             = var.engine_version
  host_instance_type         = var.host_instance_type
  deployment_mode            = "CLUSTER_MULTI_AZ"
  subnet_ids                 = var.subnet_ids
  auto_minor_version_upgrade = true

  user {
    username = var.username
    password = var.password
  }
}