Hello team, working on OAK-1717[0] and trying to reproduce the issue in a systematic way for the fix to come. So far I was not able to reproduce it. See [1] for the initial code.
[0] https://issues.apache.org/jira/browse/OAK-1717 [1] http://goo.gl/Pt5lfN By looking at log statements printed during the tests I can see the JVM is executing a quite bunch of commits per thread reducing the concurrency. Or it could be SLF4J playing tricks. I have the following ideas around it A) you identify something wrong in the code. Hooray! :) B) Try using a fixed ThreadPool[2] hoping it will increase the concurrency level C) go into a complex implementation where using wait() and notify() I control the concurrency forcing a "swap" of threads every X nodes. D) Increase the number of nodes/threads and see if it does the trick. E) reducing the `COUNT` part committing more frequently (currently 10). WDYT? D. (2) http://docs.oracle.com/javase/6/docs/api/java/util/concurrent/Executors.html#newFixedThreadPool%28int%29
