Clustering

Note: Skip this part of this documentation if you are not going to run a clustered RabbitMQ environment.

To create a cluster of RabbitMQ servers for performance and high availability, follow the following steps in addition to the previous steps on all servers.

Environment Variables

Add the following system environment variables (Start > Settings > Control Panel > System > Advanced > Environment Variables):

For server communication within the cluster we need to set the ERLANG cookie value to the same value on ALL servers that will join the cluster:

Name	Default value	Suggested value
RABBITMQ_SERVER_ADDTITIONAL_ERL_ARGS	None	-setCookie ‘xxxxx’

Important
This cookie value is an arbitrary string value that must be the same on both variables on all the servers that will join the cluster.

Erlang Cookie File

RabbitMQ service will read this erlang cookie value from a file that is stored in

C:\windows\.erlang.cookie

If there is no file present, RabbitMQ will generate a file with an arbitrary string. This file is set to read only by default. Unticking the read-only option in the file properties will allow editing of this file. The text editor needs administrator privileges to edit this file.

Set the contents of the “.erlang.cookie” file to be the same on all nodes you want in the cluster, as well as in the environment variables from previous step.

Restart the RabbitMQ service to pick up any changes to the erlang cookie.

Joining the Cluster

Run the following command and note down the nodename on the first server

rabbitmqctl status

This command will return the following

Status of node 'rabbit@123456-SERVER1' ...

Log on to all the other servers that you want jo join into a cluster and run these commands on all of them:

rabbitmqctl stop_app

rabbitmqctl join_cluster ‘rabbit@123456-SERVER1'

rabbitmqctl start_app

Verifying the Configuration

Note: The RabbitMQ service must be running at this point, and you must have enabled the management GUI by creating the file called enabled_plugins.

Login to the web management GUI interface at http://localhost:15672 on any of the servers. Log in with default credentials:

Username: guest

Password: guest

Note: The guest user can only login from localhost.

Once you login you will see all your cluster nodes listed:

Figure 1 - Example of the list of cluster nodes