MQ RabbitMQ Brokers Should Use Cluster Deployment Mode

Overview

This check verifies that your Amazon MQ RabbitMQ brokers use cluster deployment mode (CLUSTER_MULTI_AZ) instead of single-instance deployment. Cluster mode distributes broker nodes across multiple Availability Zones, providing high availability and fault tolerance.

Risk

Running a RabbitMQ broker as a single instance creates a single point of failure:

• If the instance or Availability Zone experiences an outage, your message queue becomes unavailable
• Messages in flight could be lost during unexpected failures
• Applications depending on the broker may experience significant downtime
• Message ordering and delivery guarantees may be compromised

Cluster deployment eliminates these risks by maintaining broker availability even when individual nodes or AZs fail.

Remediation Steps

Prerequisites

You need permission to create or modify Amazon MQ brokers. If you have existing single-instance brokers, you will need to create new cluster-mode brokers and migrate your applications (deployment mode cannot be changed after creation).

AWS Console Method

Important: You cannot convert an existing single-instance broker to cluster mode. You must create a new broker with cluster deployment and migrate your applications.

1. Open the Amazon MQ console at https://console.aws.amazon.com/amazon-mq
2. Click Create brokers
3. Select RabbitMQ as the engine type, then click Next
4. For Deployment mode, select Cluster deployment
5. Enter a Broker name (e.g., my-rabbitmq-cluster)
6. Choose an appropriate Broker instance type (e.g., mq.m5.large)
7. Configure RabbitMQ access:
   • Enter an Admin username
   • Enter a Password (at least 12 characters, with 4 unique characters)
8. Under Network and security:
   • Choose whether to make the broker publicly accessible
   • Select your VPC and subnets (for private brokers, select subnets in different AZs)
   • Select or create security groups
9. Click Create broker

After the new broker is active (this takes several minutes), update your applications to use the new broker endpoint, then delete the old single-instance broker.

AWS CLI (optional)

Create a new RabbitMQ broker in cluster deployment mode:

aws mq create-broker \
  --broker-name my-rabbitmq-cluster \
  --engine-type RABBITMQ \
  --engine-version 3.13 \
  --deployment-mode CLUSTER_MULTI_AZ \
  --host-instance-type mq.m5.large \
  --no-publicly-accessible \
  --auto-minor-version-upgrade \
  --subnet-ids subnet-xxxxxxxx subnet-yyyyyyyy \
  --security-groups sg-xxxxxxxx \
  --users '[{"Username":"admin","Password":"YourSecurePassword123!"}]' \
  --region us-east-1

Replace:

• subnet-xxxxxxxx and subnet-yyyyyyyy with your subnet IDs (in different AZs)
• sg-xxxxxxxx with your security group ID
• YourSecurePassword123! with a secure password

Check the broker status:

aws mq describe-broker \
  --broker-id <broker-id> \
  --region us-east-1 \
  --query '{Name:BrokerName,DeploymentMode:DeploymentMode,State:BrokerState}'

List all brokers and their deployment modes:

aws mq list-brokers \
  --region us-east-1 \
  --query 'BrokerSummaries[?EngineType==`RABBITMQ`].{Name:BrokerName,Id:BrokerId,DeploymentMode:DeploymentMode}'

CloudFormation (optional)

Use this template to create a RabbitMQ broker in cluster deployment mode:

AWSTemplateFormatVersion: '2010-09-09'
Description: Amazon MQ RabbitMQ broker in cluster deployment mode

Parameters:
  BrokerName:
    Type: String
    Description: Name for the RabbitMQ broker
    Default: my-rabbitmq-cluster

  HostInstanceType:
    Type: String
    Description: Instance type for broker nodes
    Default: mq.m5.large
    AllowedValues:
      - mq.m5.large
      - mq.m5.xlarge
      - mq.m5.2xlarge
      - mq.m5.4xlarge

  Username:
    Type: String
    Description: Admin username for the broker
    MinLength: 2
    MaxLength: 100

  Password:
    Type: String
    Description: Admin password for the broker
    NoEcho: true
    MinLength: 12

  SubnetIds:
    Type: List<AWS::EC2::Subnet::Id>
    Description: Subnet IDs for the broker (provide subnets in different AZs)

  SecurityGroupIds:
    Type: List<AWS::EC2::SecurityGroup::Id>
    Description: Security group IDs for the broker

Resources:
  RabbitMQBroker:
    Type: AWS::AmazonMQ::Broker
    Properties:
      BrokerName: !Ref BrokerName
      EngineType: RABBITMQ
      EngineVersion: '3.13'
      HostInstanceType: !Ref HostInstanceType
      DeploymentMode: CLUSTER_MULTI_AZ
      PubliclyAccessible: false
      AutoMinorVersionUpgrade: true
      SubnetIds: !Ref SubnetIds
      SecurityGroups: !Ref SecurityGroupIds
      Users:
        - Username: !Ref Username
          Password: !Ref Password

Outputs:
  BrokerArn:
    Description: ARN of the RabbitMQ broker
    Value: !GetAtt RabbitMQBroker.Arn

  BrokerEndpoints:
    Description: AMQP endpoints for the broker
    Value: !Join [',', !GetAtt RabbitMQBroker.AmqpEndpoints]

Deploy the stack:

aws cloudformation create-stack \
  --stack-name rabbitmq-cluster \
  --template-body file://template.yaml \
  --parameters \
    ParameterKey=Username,ParameterValue=admin \
    ParameterKey=Password,ParameterValue=YourSecurePassword123! \
    ParameterKey=SubnetIds,ParameterValue="subnet-xxxxxxxx,subnet-yyyyyyyy" \
    ParameterKey=SecurityGroupIds,ParameterValue=sg-xxxxxxxx \
  --region us-east-1

Terraform (optional)

variable "broker_name" {
  description = "Name for the RabbitMQ broker"
  type        = string
  default     = "my-rabbitmq-cluster"
}

variable "host_instance_type" {
  description = "Instance type for broker nodes"
  type        = string
  default     = "mq.m5.large"
}

variable "admin_username" {
  description = "Admin username for the broker"
  type        = string
}

variable "admin_password" {
  description = "Admin password for the broker"
  type        = string
  sensitive   = true
}

variable "subnet_ids" {
  description = "Subnet IDs for the broker (in different AZs)"
  type        = list(string)
}

variable "security_group_ids" {
  description = "Security group IDs for the broker"
  type        = list(string)
}

resource "aws_mq_broker" "rabbitmq_cluster" {
  broker_name        = var.broker_name
  engine_type        = "RabbitMQ"
  engine_version     = "3.13"
  host_instance_type = var.host_instance_type
  deployment_mode    = "CLUSTER_MULTI_AZ"
  publicly_accessible = false
  auto_minor_version_upgrade = true

  subnet_ids         = var.subnet_ids
  security_groups    = var.security_group_ids

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

  tags = {
    Environment = "production"
  }
}

output "broker_arn" {
  description = "ARN of the RabbitMQ broker"
  value       = aws_mq_broker.rabbitmq_cluster.arn
}

output "broker_endpoints" {
  description = "AMQP endpoints for the broker"
  value       = aws_mq_broker.rabbitmq_cluster.instances[*].endpoints
}

Create a terraform.tfvars file with your values:

admin_username     = "admin"
admin_password     = "YourSecurePassword123!"
subnet_ids         = ["subnet-xxxxxxxx", "subnet-yyyyyyyy"]
security_group_ids = ["sg-xxxxxxxx"]

Verification

After creating your cluster-mode broker, verify it is properly configured:

1. Go to the Amazon MQ console
2. Click on your broker name
3. On the Details page, confirm Deployment mode shows Cluster
4. Check that the Status shows Running

CLI verification

aws mq describe-broker \
  --broker-id <broker-id> \
  --region us-east-1 \
  --query '{Name:BrokerName,DeploymentMode:DeploymentMode,State:BrokerState}'

The output should show "DeploymentMode": "CLUSTER_MULTI_AZ".

To find brokers that are not using cluster mode:

aws mq list-brokers \
  --region us-east-1 \
  --query 'BrokerSummaries[?EngineType==`RABBITMQ` && DeploymentMode!=`CLUSTER_MULTI_AZ`].{Name:BrokerName,Id:BrokerId,DeploymentMode:DeploymentMode}'

Additional Resources

• AWS Documentation: Amazon MQ for RabbitMQ
• AWS Documentation: RabbitMQ Broker Architecture
• AWS Documentation: Amazon MQ Best Practices

Notes

• Migration required: You cannot change the deployment mode of an existing broker. You must create a new cluster-mode broker and migrate your applications.
• Cost considerations: Cluster deployment uses multiple broker instances across AZs, which increases costs compared to single-instance deployment. However, the improved availability typically justifies the additional cost for production workloads.
• Instance type requirements: Cluster deployment requires at least mq.m5.large instance type.
• Network configuration: For private cluster brokers, you need subnets in at least two different Availability Zones.
• Client reconnection: Ensure your applications implement automatic reconnection logic with exponential backoff to handle node failovers gracefully.