Chapter 1. Introduction
Red Hat AMQ Broker 7.5 is available as a containerized image that is provided for use with OpenShift Container Platform (OCP) 3.11 and later.
AMQ Broker is based on Apache ActiveMQ Artemis. It provides a message broker that is JMS-compliant. After you have set up the initial broker pod, you can quickly deploy duplicates by using OpenShift Container Platform features.
AMQ Broker on OCP provides similar functionality to Red Hat AMQ Broker, but some aspects of the functionality need to be configured specifically for use with OpenShift Container Platform.
1.1. Version compatibility and support
For details about OpenShift Container Platform 4.1 image version compatibility, see the OpenShift and Atomic Platform Tested Integrations page.
1.2. Unsupported features
Master-slave-based high availability
High availability (HA) achieved by configuring master and slave pairs is not supported. Instead, when pods are scaled down, HA is provided in OpenShift by using the scaledown controller, which enables message migration.
External Clients that connect to a cluster of brokers, either through the OpenShift proxy or by using bind ports, may need to be configured for HA accordingly. In a clustered scenario, a broker will inform certain clients of the addresses of all the broker’s host and port information. Since these are only accessible internally, certain client features either will not work or will need to be disabled.
Client Configuration Core JMS Client
Because external Core Protocol JMS clients do not support HA or any type of failover, the connection factories must be configured with
useTopologyForLoadBalancing=false.AMQP Clients
AMQP clients do not support failover lists
Durable subscriptions in a cluster
When a durable subscription is created, this is represented as a durable queue on the broker to which a client has connected. When a cluster is running within OpenShift the client does not know on which broker the durable subscription queue has been created. If the subscription is durable and the client reconnects there is currently no method for the load balancer to reconnect it to the same node. When this happens, it is possible that the client will connect to a different broker and create a duplicate subscription queue. For this reason, using durable subscriptions with a cluster of brokers is not recommended.
