Closed scottmarlow closed 8 years ago
N.B. that Hibernate OGM generally expects to use the latest version of Infinispan for purposes of NoSQL storage, so we actively avoid using the version within WildFly (slot "main") but rather expect users to download the "WildFly/EAP modules" from http://infinispan.org/download/
Enabling 2nd level cache should of course use the existing code, and therefore rely on the slot "main" of Infinispan for caching purposes.
So if Hibernate OGM can use the same version of Hibernate ORM as provided within WildFly, there should be no need for a different module of Hibernate ORM. The existing module is already setup to load those services and there should be no need for changes in clustering setup either.
Having help from the Clustering and Infinispan team would be useful to setup the custom Infinispan module, i.e. allow it to use FORK, but this is unrelated to the 2LC issue.
Is this issue modified somehow after the new Jipijapa OGM integration that @Sanne worked on?
Is this issue modified somehow after the new Jipijapa OGM integration that @Sanne worked on?
Good point, I think it would work now in an equivalent way as ORM works. We didn't test this though.
Created an issue on OGM to test for this:
Could we consider this issue resolved?
Yes, I will close this issue as it now duplicates OGM-1077, thanks!
Hibernate OGM on WildFly, may have a copy of the Hibernate ORM jars that OGM depends on. Some potential issues with having different copies of ORM in WildFly:
Perhaps the default WildFly OGM module, will depend on the same ORM version but not the same WildFly ORM module. If we want OGM to be able to work with a clustered 2lc, via the WildFly Infinispan module, that has to be enabled somehow to work for each deployment that uses OGM. This might require some breaking Hibernate ORM changes, to better integrate the WildFly ORM regionFactory extensions (may also be nicer if WildFly didn't need to extend the Hibernate-Infinispan RegionFactory classes).