Configure RabbitMQ failover cluster

RabbitMQ is a universal bus to be used to exchange messages between the applications. There are other solutions like Kafka, AWS SQS, Google Pubsub, etc. If you are here then I suppose you already know what RabbitMQ is and I will not go deep into details.

This article describes the installation and configuration of the RabbitMQ in CentOS v7.

The key trick in RabbitMQ is in fact that there is no such thing as global cluster. The master role is being dedicated to one of the ervers within the scope of a single event in the queue. The outage of one server doesn’t cuse the downtime and you don’t need to care about doing something manually when it is back.

Nex’t I’ll provide step by step instructions how to configure failover cluster for rabbitMQ that consists of 3 servers.

The most officient way to do this is to run it in Docker containers.

Run on all servers

So you got to have docker and docker-compose installed on all 3 servers

  1. Docker install instructions for Ubuntu
  2. Docker install instructions for Centos
  3. docker-compose install instructions for Linux on all the servers.

RabbitMQ cluster is using the erlang cookie to authorize nodes in the cluster. You can generate any random string for this.

Run the following on all servers:

mkdir -p /opt/rabbitmq/{data,logs}
echo ${ERLANG_COOKIE} > /opt/rabbitmq/data/.erlang.cookie

chmod 600 /opt/rabbitmq/data/.erlang.cookie
chmod 777 /opt/rabbitmq/log/

Servers inside the cluster need to be able to resolve each other with the short DNS names. For this you’ll need to update the /etc/hosts file on all servers:

echo """ server-1 server-2 server-3
""" >> /etc/hosts

Create docker-compose.yaml

cat <<EOF > /opt/rabbitmq/docker-compose.yaml
version: '2'
    image: rabbitmq:3.10-management
    container_name: '$(uname -n)'
        - RABBITMQ_NODENAME=rabbit@$(uname -n)
        - 4369:4369
        - 5672:5672
        - 15672:15672
        - 25672:25672
        - /opt/rabbitmq/data/:/var/lib/rabbitmq/
        - /opt/rabbitmq/log/:/var/log/rabbitmq

Start containers:

docker-compose -f /opt/rabbitmq/docker-compose.yaml up -d 

Connect all servers into a cluster

The following commands should be executed on the slave nodes. You can use server-1 as master and run the following on server-2 and server-3:

docker exec --tty $(uname -n) rabbitmqctl stop_app
docker exec --tty $(uname -n) rabbitmqctl reset
docker exec --tty $(uname -n) rabbitmqctl join_cluster rabbit@server-1
docker exec --tty $(uname -n) rabbitmqctl start_app

The following can be used to get the data from the cli:

rabbitmqctl cluster_status

Sample output:

docker exec --tty $(uname -n) rabbitmqctl cluster_status
Cluster status of node rabbit@cd27271cd9ff ...


RabbitMQ has a nice web UI that can be very helpfull in every day usage. It is disabled by default. Run the following in order to enabpe it:

rabbitmq-plugins enable rabbitmq_management

RabbitMQ will prompt to be restarted and after it the UI will be available on the 15672 port. Use the server IP address to access it http://%server_ip_address%:15672/

Standard user:

  • login: guest
  • password: guest

On the first page you’ll see the state of the cluster (there is only one server in it so far)

You can use HA-Proxy to distribute the load between the servers i the cluster

The server-sde of the configuration is finished at this step

Examples to create users, hosts and queues

You can use cli to create users, hosts and message queues. Since all the commands are to be executed in bash you can create a set of variables that will repeat from server to server


The following commands can be executed on any node in the cluster inside the docker container:

docker exec --tty $(uname -n) 

Or attach to the container console by running the following:

docker exec -it $(uname -n) bash

Create vhost:

rabbitmqadmin declare vhost name=${vhost}

Create a user to access the vhost

rabbitmqadmin declare user name=${user} password=${password} tags=administrator  
rabbitmqadmin declare permission vhost=${vhost} user=${user} configure=`.\*` write=`.\*` read=`.*`

Create exchange buffer:

rabbitmqadmin -V ${vhost} -u ${user} -p ${password} declare exchange name=${exchange} type=fanout

Create message queue:

rabbitmqadmin -V ${vhost} -u ${user} -p ${password} declare queue name=${queue_name} durable=true

Create exchange buffer and a queue:

rabbitmqadmin -V ${vhost} -u ${user} -p ${password} declare binding source=${exchange} destination_type=queue destination=${queue_name}

Define HA policies for the cluster:

rabbitmqadmin -V ${vhost} -u ${user} -p ${password} declare policy name=ha_all pattern=`.*` definition='{`ha-mode`:`all`,`ha-sync-mode`:`automatic`}' apply-to=all

Usefull links: