Can you give me more of the stack trace? This is on a Spark executor? Wild guess is that you need more memory available to the JVM on the executor machine.
On May 24, 2017, at 1:21 AM, Dennis Honders <dennishond...@gmail.com> wrote: I receive the following error when training with UR 0.6.0 master branche. [ERROR] [Executor] Managed memory leak detected; size = 80424024 bytes, TID = 26 It does not kill the process or anything like that (able to deploy and query), but maybe needs some attention.