The fix for this includes both upgrading our remoting jar (from 3.2.2 to 3.2.7, a bug-fix only update), and properly interrupting the close() operation after 5 seconds of waiting. This ensures The threads clean up after themselves.