[
https://issues.apache.org/jira/browse/ARTEMIS-3640?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17482432#comment-17482432
]
Gary Tully edited comment on ARTEMIS-3640 at 1/26/22, 4:43 PM:
---------------------------------------------------------------
I was thinking along the lines of two cluster connections (one that is
configured to do very little but with the external dns conector-ref) and having
an external and internal acceptor configured, the external acceptor linking to
the 'external' connector ref. I don't think this will hang together b/c the
topology internally is broker wide, and not tied to an acceptor, despite the
possibility that an acceptor can link to a specific cluster connection. Also, I
don't think there is a way to separate topology from clustering data on a
cluster connection at the moment.
However, peeking more and after having a chat with [~andytaylor], the
connector-ref may still be the answer.
I wonder if it would be possible to have the cluster connector-ref reference a
uri with both the internal and external dns names?
The url format with multiple hosts is a little strange, I found an example at:
https://github.com/apache/activemq-artemis/blob/main/artemis-jms-client/src/test/java/org/apache/activemq/artemis/uri/ConnectionFactoryURITest.java#L233
{code:java}
<connectors>
<connector
name="artemis-master">tcp://internal-cluster-dns-1:61616</connector>
<connector
name="artemis-slave">tcp://internal-cluster-dns-2:61616</connector>
<!-- expose external and internal dns to clients - reference this from the
cluster-connecton.connector-ref may need to verify the exact format! -->
<connector
name="artemis-master-internal-and-external">(tcp://internal-cluster-dns-1:61616?,tcp://external-cluster-dns-1:61616?)</connector>
</connectors>
{code}
and share that as topology information, to both internal and external clients.
{code:java}
<cluster-connections>
<cluster-connection name=...>
....
<connector-ref>artemis-master-internal-and-external</connector-ref>
<static-connectors>
<connector-ref>artemis-slave</connector-ref>
</static-connectors>
</cluster-connection>
</cluster-connections>
{code}
was (Author: gtully):
I was thinking along the lines of two cluster connections (one that is
configured to do very little but with the external dns conector-ref) and having
an external and internal acceptor configured, the external acceptor linking to
the 'external' connector ref. I don't think this will hang together b/c the
topology internally is broker wide, and not tied to an acceptor, despite the
possibility that an acceptor can link to a specific cluster connection. Also,
I don't think there is a way to separate topology from clustering data on a
cluster connection at the moment.
However, peeking more and after having a chat with [~andytaylor], the
connector-ref may still be the answer.
I wonder if it would be possible to have the cluster connector-ref reference a
failover uri with both the internal and external dns names?
{code}
<connectors>
<connector
name="artemis-master">tcp://internal-cluster-dns-1:61616</connector>
<connector
name="artemis-slave">tcp://internal-cluster-dns-2:61616</connector>
<!-- expose external and internal dns to clients - reference this from the
cluster-connecton.connector-ref -->
<connector
name="artemis-master-internal-and-external">failover:(tcp://internal-cluster-dns-1:61616,tcp://external-cluster-dns-1:61616)</connector>
</connectors>
{code}
and share that as topology information, to both internal and external clients.
{code}
<cluster-connections>
<cluster-connection name=...>
....
<connector-ref>artemis-master-internal-and-external</connector-ref>
<static-connectors>
<connector-ref>artemis-slave</connector-ref>
</static-connectors>
</cluster-connection>
</cluster-connections>
{code}
> external clients cannot use cluster topology for HA or load balancing
> ---------------------------------------------------------------------
>
> Key: ARTEMIS-3640
> URL: https://issues.apache.org/jira/browse/ARTEMIS-3640
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: Vilius Šumskas
> Assignee: Gary Tully
> Priority: Major
>
> Let's say there is a simple Artemis HA master/slave shared storage cluster
> with such configuration:
>
> {noformat}
> <connectors>
> <connector
> name="artemis-master">tcp://internal-cluster-dns-1:61616</connector>
> <connector
> name="artemis-slave">tcp://internal-cluster-dns-2:61616</connector>
> </connectors>
> {noformat}
> The cluster is using static discovery.
>
> Factory on internal clients are using:
> {noformat}
> (tcp://internal-cluster-dns-1:61616,tcp://internal-cluster-dns-2:61616)?ha=true&reconnectAttempts=30&useTopologyForLoadBalancing=false{noformat}
> Those internal clients can successfully connect to the cluster and failover
> works as expected.
>
> However if there is also external clients which use external DNS/IP address
> of the cluster the topology doesn't return correct DNS name and the failover
> fails.
> {noformat}
> (tcp://external-cluster-dns-1:61616,tcp://external-cluster-dns-2:61616)?ha=true&reconnectAttempts=30&useTopologyForLoadBalancing=false{noformat}
> {noformat}
> org.apache.activemq.artemis.core.remoting.impl.netty.NettyConnector:
> AMQ214033: Cannot resolve host java.net.UnknownHostException:
> internal-cluster-dns-1{noformat}
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
