[jira] [Created] (ZEPPELIN-4481) Production running

["Verbeque (Jira)" <ji...@apache.org>]

Verbeque created ZEPPELIN-4481:
----------------------------------

             Summary: Production running
                 Key: ZEPPELIN-4481
                 URL: https://issues.apache.org/jira/browse/ZEPPELIN-4481
             Project: Zeppelin
          Issue Type: Bug
    Affects Versions: 0.8.0
            Reporter: Verbeque


Hi all,

I'm here to report multiple bugs that I encountered while running Zeppelin inside a kerberized environment.
 * I confirm https://issues.apache.org/jira/browse/ZEPPELIN-1770 isn't resolved at all
 * I confirm https://issues.apache.org/jira/browse/ZEPPELIN-4045
 * After few hours zeppelin master process will spawn many threads (14000+ after raising the limit) and reach default nproc limit (4096) and report errors unrelated to the unability to launch new threads (thrift error...), I think it would be great to document it.
 * When reaching 100+ notebooks the server will raise an OOM and some notebooks will disappear from the UI, I don't know if you have any recommendations for xmx value ?
 * Spark interpreter and pyspark keeps dying randomly. It would be great to track pids of this process (in order to lazily restart them if needed ?).
 * After few hours spark interpreter loose spark.sql.hive.hiveserver2.jdbc.url and spark.sql.hive.hiveserver2.jdbc.url.principal configuration parameters.

By the way Zeppelin is great product and the UI is really nice, thanks for the hard work!



--
This message was sent by Atlassian Jira
(v8.3.4#803005)