multi-host jms cluster on wildfly10.0.0?

[emitani]

Hello,

I'm currently trying to set up a jms cluster with full-ha profile.

I got to the point where jms cluster works fine if I have two nodes in the same machine. When I send a topic message, both instances receive it, which is what I want.

But when I try to set up a cluster across two machines, it seems to me like each machine has its own cluster, and they don't know about each other.

Currently I use default settings for messaging-activemq subsystem in domain.xml, other than changing cluster user and password. Both hosts have the exact same copy of domain.xml. 

How can I configure a cluster across multiple hosts? Am I missing some additional configuration?

Below is the output from a jboss-cli command. Each node has two active server instances. maser has ip=10.10.35.59, and slave ip is 10.10.35.38:

[domain@w7-0476:9990 cluster-connection=my-cluster] /host=master/server=server-one/subsystem=messaging-activemq/server=default/cluster-connection=my-cluster:read-attribute(name=topology)

{

    "outcome" => "success",

    "result" => "topology on Topology@6d5d458e[owner=ClusterConnectionImpl@1727559813[nodeUUID=1b568e94-051c-11e7-9631-57d6194264c7, connector=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8080&host=10-10-35-59, address=jms, server=ActiveMQServerImpl::serverUUID=1b568e94-051c-11e7-9631-57d6194264c7]]:

        1d6b651a-051c-11e7-a85c-217e75985f77 => TopologyMember[id = 1d6b651a-051c-11e7-a85c-217e75985f77, connector=Pair[a=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8230&host=10-10-35-59, b=null], backupGroupName=null, scaleDownGroupName=null]

        1b568e94-051c-11e7-9631-57d6194264c7 => TopologyMember[id = 1b568e94-051c-11e7-9631-57d6194264c7, connector=Pair[a=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8080&host=10-10-35-59, b=null], backupGroupName=null, scaleDownGroupName=null]

        nodes=2 members=2"

}

[domain@w7-0476:9990 cluster-connection=my-cluster] /host=slave/server=server-one/subsystem=messaging-activemq/server=default/cluster-connection=my-cluster:read-attribute(name=topology)

{

    "outcome" => "success",

    "result" => "topology on Topology@49c21d[owner=ClusterConnectionImpl@17675183[nodeUUID=292110a7-051e-11e7-b52d-a7a7ab3e2455, connector=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8080&host=10-10-35-85, address=jms, server=ActiveMQServerImpl::serverUUID=292110a7-051e-11e7-b52d-a7a7ab3e2455]]:

        292110a7-051e-11e7-b52d-a7a7ab3e2455 => TopologyMember[id = 292110a7-051e-11e7-b52d-a7a7ab3e2455, connector=Pair[a=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8080&host=10-10-35-85, b=null], backupGroupName=null, scaleDownGroupName=null]

        2a2c669c-051e-11e7-afb9-7fdbf16f1ab2 => TopologyMember[id = 2a2c669c-051e-11e7-afb9-7fdbf16f1ab2, connector=Pair[a=TransportConfiguration(name=http-connector, factory=org-apache-activemq-artemis-core-remoting-impl-netty-NettyConnectorFactory) ?httpUpgradeEnabled=true&httpPpgradeEndpoint=http-acceptor&port=8230&host=10-10-35-85, b=null], backupGroupName=null, scaleDownGroupName=null]

        nodes=2 members=2"

}

Thank you very much for your help.

[mnovak]

I think the problem could be in socket-bindings for your profile.

Artemis is using "udp" JGroups stack by default to discover nodes in cluster. It's using "jgroups-udp" socket binding:

<socket-binding name="jgroups-udp" interface="private" port="55200" multicast-address="${jboss.default.multicast.address:230.0.0.4}" multicast-port="45688"/>

which is be default on "private" interface:

<interface name="private">
      <inet-address value="${jboss.bind.address.private:127.0.0.1}"/>
    </interface>

as you can see it binds to 127.0.0.1 so udp multicast cannot go to the other machine.

Could you modify the configuration of "jgroups-udp" socket binding to use "public" interface:

<socket-binding name="jgroups-udp" interface="public" port="55200" multicast-address="${jboss.default.multicast.address:230.0.0.4}" multicast-port="45688"/>

This is wild shot :-) I'll need to see your configuration to know better.

Thanks,

Mirek

[emitani]

Hi Mirek, that resolved the problem! My jms cluster is now working as expected. Thank you very much!

[mnovak]

You're welcome! I'm glad it helped.