Thanks Andy, I will switch to the datanucleus forum for this specific problem.
More generally, I did hope that adding caching was low-hanging fruit, but seems its more complex than I imagined. Maybe not the thing to tackle in the weeks before Christmas. My app is read intensive, so adding caching is logical, but what actually happens - by default - with addition of a JSR-107 JCACHE layer (via DataNucleus) is the question, and my task to understand and describe. That error relates to creation of a new record, reads seem fine, but are they being cached? DataNucleus seems to save on total queries by doing complex SQL joins, but can the results of such queries be cached I wonder, or is that too complex a thing to manage? Trying to log the chache activity didn't give much info, maybe nothing is happening to log. Another option is to use Neo4j, which potentially has a similar improvement on speed but does the in-memory caching of object graphs (the complexity and main bottleneck?) at the database level. Steve On Sun, Dec 11, 2016 at 12:30 AM, Andy Jefferson <[email protected]> wrote: > > However, the following exception is common to the range of cache > implementations I've tried. I'm using datanucleus-core.4.1.7.jar > > I'll see if Andy will talk to me .. > > - java.lang.ClassCastException > > - org.datanucleus.identity.SCOID cannot be cast to java.io.Serializable > > - org.datanucleus.cache.ehcache.EhcacheLevel2Cache#get( > EhcacheLevel2Cache.java:143) > > Hi Stephen, > please use the DataNucleus support process for such things; reporting them > on a third party site isn't likely to get looked at with any frequency; I > only rarely scan down ISIS mailing lists. > What you post says little about what you're doing, other than using > "nondurable identity" for some reason (which clearly will not be cacheable > since those objects have no "identity") > > > Suggest that you use either of > http://forum.datanucleus.org > https://gitter.im/datanucleus/Lobby > > Problem reporting is described on > http://www.datanucleus.org/documentation/problem_reporting.html > > > > Regards > -- > Andy > DataNucleus (Web: http://www.datanucleus.org Twitter: @datanucleus) >
