Integrate Infinispan Server Endpoint into Wildfly 10.1.0.Final

[sdyarnell]

Hi,

I am trying to expose a cache externally using Wildfly 10.1.0.Final. In my server, I am selectively replicating the underlying database manually (so not all changes are replicated everywhere). Because of this, I need to explore invalidating the cache values manually.

I would prefer to integrate the infinispan server endpoint subsystem into my current wildfly setup and configure it like below:

        <subsystem xmlns="urn:infinispan:server:endpoint:8.0">

            <hotrod-connector socket-binding="hotrod" cache-container="hibernate">

                <topology-state-transfer lazy-retrieval="false" lock-timeout="1000" replication-timeout="5000"/>

            </hotrod-connector>

        </subsystem>

Is this something I should be able to accomplish? If so, how do I go about discovering the required modules (especially the right version of those modules) to move into my current Wildfly deployment?

Final question: will I even be able to invalidate the entity cache like I am trying to do? If not, then I would rather not waste the time trying, lol.

Thanks,

Shawn

[pferraro]

If you access your database via JPA, you can use WildFly's distributed second-level cache which will do all of this for you.

[sdyarnell]

I do not understand how any piece of software in wildfly would know to do anything if I simply run a SQL statement on the underlying database. Can you please elaborate on how that is the case?

[pferraro]

I didn't say anything about running a SQL statement via JDBC.  I said that, if you use JPA for database access, you can leverage Hibernate's 2nd-level cache to perform caching/invalidation for you.

Using a remote cache to persist entities largely defeats the purpose of this level of caching, since it requires a network call to access the cached data.