Github user steveloughran commented on the issue:
https://github.com/apache/zeppelin/pull/2618
# Spark 2.3 has a spark-hadoop-cloud module which adds a set of
dependencies on the hadoop cloud modules, throwing out all the transitive
conflicting modules. Either pick that up or lift all the exclusion logic there
to keep out things like: jackson, guava, joda-time which will only cause pain.
Hadoop 2.9+ will move to a shaded aws JAR here.
Do be aware, hadoop-aws is very fussy about aws-SDK versions: that library
is a moving target. You can't expect to change versions and not expect to have
the code link, let alone run.
---
