Kundera Datastax CQL Driver - Autodiscovery of Cassandra Nodes

[dhayek]

Hi

I'm working with version 2.13.1 of the Kundera Cassandra Datastax driver:

        <groupId>com.impetus.kundera.client</groupId>
        <artifactId>kundera-cassandra-ds-driver</artifactId>
        <version>2.13.1</version>

We have three nodes in our ring, but the Kundera Datastax driver only seems to connect to the one host that is configured in persistence.xml. (I noticed this issue when that host went down briefly.)

When I tried using the Datastax CQL driver directly with the same initial host, it autodiscovered the other 2 in the ring.

What properties do I need to set in persistence.xml to force the Kundera Datastax driver to autodiscover all of the nodes in the ring?

Thanks

David

[DaVincii]

In order to make all the nodes in your ring available to Kundera, you need to specify them in an external configuration file.

For further details check this link :

https://github.com/impetus-opensource/Kundera/wiki/Using-multiple-node-support-and-load-balancing-policy-in-Kundera

[chhavigangwal]

@dhayek

Is the issue resolved at your end ?

Chhavi

[avinashkp]

Alternatively can I add the nodes dynamically using properties being set while creating the EntityManagerFactory? http://stackoverflow.com/questions/13140082/change-content-in-persistence-xml-dynamically-during-execution

[hkroger]

Hmm, DataStax driver that is being used underneath handles these a bit smarter than Kundera configuration currently allows. It would be great if those could be utilized using Kundera as well.

For example following things are missing in Kundera but available in underlying driver:

• Auto discovery of nodes (adding/removing nodes on the fly)
• Cassandra specific load balancing policies like TokenAwarePolicy (sending request to node that holds one copy of the data in question)