Hi Gang, I face some issues using the Native rdf indexing service within a multi-threaded context.
I have no code to share ATM that demonstrate the issue but if it is needed I'll be able to produce and share it next week. Let me explain my setup first : - my app is running in a container - on every http request I assemble and activate a qi4j app, process the request and then passivate the qi4j app - some base data is created in the qi4j app (entities inserted, indexed) After that I have a stress test that make concurrent requests where the processing do a rdf query on base data and create new entities. Base data inserted before the test is not changed. The issue is that concurrent requests don't get the same query results from the base data. I tried to change the version of the openrdf projects used by qi4j from 2.3.0/2.3.1 to 2.3.2 everywhere and rerun my stress test. With this version I got a java.nio.channels.OverlappingFileLockException from a SailRepository class. I then digged in the OpenRDF forums where I found a thread titled "Concurrent access using Native RDF Repository" [1] that got me disapointed. How do you use indexing-rdf in multi-threaded contexts ? /Paul [1] http://www.openrdf.org/forum/mvnforum/viewthread?thread=755 _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
