Marius Krämer created ZEPPELIN-5108:
---------------------------------------
Summary: Unnecessary check for existance of `hive-site.xml` for
hive support of spark interpreter
Key: ZEPPELIN-5108
URL: https://issues.apache.org/jira/browse/ZEPPELIN-5108
Project: Zeppelin
Issue Type: Improvement
Components: Interpreters
Affects Versions: 0.8.2
Reporter: Marius Krämer
There's a potentially unnecessary gotcha when trying to activate hive support
of zeppelins spark interpreter: In case no hive-site.xml is found, support
won't be be enabled - even though a hive-site.xml might no be necessary for a
given setup.
We're using a spark / delta lake setup without any hive components besides an
external metastore, and all hive-related config is being done in
spark-defaults.conf via sparks spark.hadoop.xxx settings, e.g.
{code:java}
spark.hadoop.hive.metastore.uris thrift://hive-metastore:9083{code}
To enable hive support in zeppelins spark interpreter, we have to create
hive-site.xml which AFAICS is functionally just not necessary. I realize that
the need for an hive-site.xml is actually documented, but it took me a while to
figure out that there's really a hardcoded file check being done here :)
Would there be any downside to dropping the hiveSiteExisted check
[here|https://github.com/apache/zeppelin/blob/4c02b75d8e7b3b5f2a1a5af8bef051049fbc5766/spark/spark-scala-parent/src/main/scala/org/apache/zeppelin/spark/BaseSparkScalaInterpreter.scala#L286]?
I'd by happy to try and contribute if this makes sense to you :) Thanks for
Zeppelin btw, it's awesome.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
