You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zeppelin.apache.org by "Ashish K Dubey (JIRA)" <ji...@apache.org> on 2018/01/21 03:03:00 UTC

[jira] [Created] (ZEPPELIN-3178) JDBCInterpreter creates new connection/session for individual paragraph

Ashish K Dubey created ZEPPELIN-3178:
----------------------------------------

             Summary: JDBCInterpreter creates new connection/session for individual paragraph
                 Key: ZEPPELIN-3178
                 URL: https://issues.apache.org/jira/browse/ZEPPELIN-3178
             Project: Zeppelin
          Issue Type: Improvement
          Components: Interpreters
            Reporter: Ashish K Dubey


Within one notebook when i run different SQL in different paragraphs - it maintains a hashmap of connection object per paragraph id ( did check in 0.6 ) and in brief look at 0.7 also this seems like the case - now the side effect of the same is - some databases like postgres , hive etc. support temporary table like object which are tied with individual session. Temp table object is really great for storing intermediate data in fast access layer ( in hive it creates table on HDFS while your main data might be in cloud storage), but due to per para session - i cannot use it across different paragraphs within one notebook which is really painful because technically most times one user owns the notebook and would want such control at the cost of parallelism .. I think this can be solved ( tested in 0.6 code base ) by maintaining connection map at noteId level . This can be done behind a parameter which can be set at interpreter setting so that default behavior remains the same.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)