Two clusters on same network

[ejb3workshop]

I would like to run two (or more) hornetq clusters on the same network. Initially I was hoping that simply having different data directories would be sufficient, however this does not seem to be the case as messages were distributed across all servers. I then changed the cluster-connection name on one of the clusters from mycluster to my-other-cluster

<cluster-connection name="my-other-cluster">

However messages were still being distributed between the clusters.  Only once I changed the discovery-group name did message remain on their respective cluster.

<discovery-groups>

    <discovery-group name="my-other-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-other-cluster">

      <address>jms</address>

      <connector-ref>netty</connector-ref>

      <retry-interval>500</retry-interval>

      <use-duplicate-detection>true</use-duplicate-detection>

      <forward-when-no-consumers>false</forward-when-no-consumers>

      <discovery-group-ref discovery-group-name="my-other-group1"/>

    </cluster-connection>

  </cluster-connections>

This suggests that the discovery-group name is sufficient to distinguish between clusters. I want to confirm that this is the case and that this is the minimum amount of change to the configuration to ensure each cluster can run independenly of each other, while still being able to discover and access nodes within the same cluster.

[ataylor]

No, the name is just used for other services to look up the discovery group. You will need to change either the address or port for both the broadcast and the discovery groups

[ejb3workshop]

That is what I was affraid of. However it seems to be working at the moment with only a different name, which is rather strange. Is there an explanation for this ?

[ataylor]

That is what I was affraid of. However it seems to be working at the moment with only a different name, which is rather strange. Is there an explanation for this ?

Nope, apart from coincedence. Just change the address and port like i suggested to gain the desired effects you want

[ejb3workshop]

And I only have to change this on the discovery group ?

[ataylor]

yes, just use a unique address for each cluster

[zivley]

Hi there,

My need is exactly what Alexander didn't want, I want to have two clusters where messages are actually distributing across both of them, and I wanted to make sure if the solution is so simple as just setting the same IP or port on both discovery groups and that's it? Is that a correct way to do it or there is a better way?

I wouldn't want to base my solution on something that works "by mistake"

Thanks in advance

[jbertram]

Nodes which use discovery groups find each other based on the use of a shared group address and port.  Therefore, if two nodes share the same discovery address and port then they should find each other granted the network allows UDP multi-cast traffic between them.