Currently the root cause for a failure goes through a bunch of additional exceptions inside the Java SDK harness.
Consider reporting just the users stack trace if the exception happened within user code instead of obfuscating it with other exceptions while the stack is unwinding.
Imported from Jira BEAM-7584. Original Jira may contain additional context.
Reported by: lcwik.