Right configuration for HA cluster ActiveMQ Artemis

Question

I'm new to ActiveMQ Artemis and ask community to check if I am right in configuration of HA cluster of brokers or may be I should configure them in another way as I haven't found detailed tutorial on my case. All of the brokers run on the same machine.

The scenario:

There is a master node on 61617 port and two slave nodes (slave1, slave2) on ports 61618 and 61619. If master node dies, one of slaves become active (replication mode).

It's necessary for the consumer to communicate with cluster as a "black-box". By that I mean that the change of master (i.e. when master dies) shouldn't have any effect on consumer (i.e. the way it connects to the cluster).

What I managed to do (as I understand for this case we should configure only cluster, acceptor, and connector properties, thus I attach only this part of configuration of brokers):

master broker:

<connectors>
    <connector name="artemis">tcp://localhost:61617</connector> 
</connectors>

<ha-policy>
    <replication>
        <master/>
    </replication>   
</ha-policy>

<acceptors>
    <acceptor name="artemis">tcp://localhost:61617</acceptor>
</acceptors>

<cluster-user>cluster</cluster-user>
<cluster-password>cluster</cluster-password>
<broadcast-groups>
    <broadcast-group name="bg-group1">
        <group-address>231.7.7.7</group-address>
        <group-port>9876</group-port>
        <broadcast-period>5000</broadcast-period>
        <connector-ref>artemis</connector-ref>
    </broadcast-group>
</broadcast-groups>
<discovery-groups>
    <discovery-group name="dg-group1">
        <group-address>231.7.7.7</group-address>
        <group-port>9876</group-port>
        <refresh-timeout>10000</refresh-timeout>
    </discovery-group>
</discovery-groups>
<cluster-connections>
    <cluster-connection name="my-cluster">
        <connector-ref>artemis</connector-ref>
        <message-load-balancing>ON_DEMAND</message-load-balancing>
        <max-hops>0</max-hops>
        <discovery-group-ref discovery-group-name="dg-group1"/>
    </cluster-connection>
</cluster-connections>

slave 1 broker the cluster conf is the same with master (auto-configuration when creating a node through the console --clustered)

<ha-policy>
    <replication>
        <slave/>
    </replication>
</ha-policy>

<connectors>
    <connector name="artemis">tcp://localhost:61618</connector>
    <connector name="netty-live-connector">tcp://localhost:61617</connector>
</connectors>

<acceptors>
    <acceptor name="artemis">tcp://localhost:61618 </acceptor>
</acceptors>

slave 2 broker the cluster conf is the same with master (auto-configuration when creating a node through the console --clustered)

<ha-policy>
    <replication>
        <slave/>
    </replication>
</ha-policy>

<connectors>
    <connector name="artemis">tcp://localhost:61619</connector>
    <connector name="netty-live-connector">tcp://localhost:61617</connector>
</connectors>

<acceptors>
    <acceptor name="artemis">tcp://localhost:61619</acceptor>
</acceptors>

JNDI configuration in consumer :

java.naming.factory.initial=org.apache.activemq.artemis.jndi.ActiveMQInitialContextFactory
connectionFactory.ConnectionFactory=(tcp://localhost:61617?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10,tcp://localhost:61618?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10,tcp://localhost:61619?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10)

My configuration works, however I don`t sure if it is the right way it should be.

I've also found similar question which uses static connectors. What are they doing? I don't understand how they work. Or may be that is the right way of configuration that I am looking for?

Justin Bertram Justin Bertram · Accepted Answer · 2019-10-29T14:29:54

The first thing to note is that using a single live/backup pair (or even live/backup/backup triplet) with network replication is dangerous due to the risk of "split-brain." I would recommend you use either 1 live/backup pair with shared-storage or 3 live/backup pairs with replication (which will allow the establishment of a proper quorum). Read the documentation about split brain for more details.

Aside from the risk of split-brain the broker configuration looks OK. Most (if not all) the configuration details are covered in the clustering and HA documentation. There is also a wealth of examples which ship with the broker many of which are specific to clustering and HA.

You could simplify your connection factory URL. Currently you have:

(tcp://localhost:61617?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10,tcp://localhost:61618?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10,tcp://localhost:61619?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10)

However, you could use:

(tcp://localhost:61617,tcp://localhost:61618,tcp://localhost:61619)?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=10

Static connectors are typically used in environments which don't support UDP multicast. It allows manual configuration of the cluster members. If you are in an environment which supports UDP multicast I recommend you use the discovery/broadcast groups configuration rather than static discovery.

In general, if everything is working the way you want that indicates your configuration is fine.

Right configuration for HA cluster ActiveMQ Artemis

1 Answers