Github user felixcheung commented on the issue:
https://github.com/apache/zeppelin/pull/2899
maybe a downloadspark script that can launch from
https://github.com/apache/zeppelin/blob/master/bin/interpreter.sh#L116?
I am not sure if it address the usability concerns fully - user can still
be confused why the spark interpreter takes so long to start? right, some sort
of messaging will be nice.---
