[ 
https://issues.apache.org/jira/browse/SPARK-6540?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Alex Liu updated SPARK-6540:
----------------------------
    Description: 
e.g.
By using beeline
> set cassandra.username=user;
> set cassandra.password = pass;
> show databases;

The "show databases" query fails due to missing cassandra.username and 
cassandra.password configuration in hiveconf when create the metastore client.

Same test passes in HiveServer2.

This document 
https://cwiki.apache.org/confluence/display/Hive/HiveServer2+Thrift+API 
mentions how HiveServer2 maps the connection to session(explicit support for 
sessions in the client API, e.g every RPC call references a session ID which 
the server then maps to persistent session state.), but Spark Hive thrift 
server fails to delivery the same function though it wraps HiveServer2 
internally.

Spark Hive thrift server contains a single hive context  which has a single 
hiveConf shared by all sessions. Ideally we should isolate it per session.

  was:
e.g.
By using beeline
> set cassandra.username=user;
> set cassandra.password = pass;
> show databases;

The "show databases" query fails due to missing cassandra.username and 
cassandra.password configuration in hiveconf when create the metastore client.

Same test passes in HiveServer2.

This document 
https://cwiki.apache.org/confluence/display/Hive/HiveServer2+Thrift+API 
mentions how HiveServer2 maps the connection to session(explicit support for 
sessions in the client API, e.g every RPC call references a session ID which 
the server then maps to persistent session state.), but Spark Hive thrift 
server fails to delivery the same function though it wraps HiveServer2 
internally.



> Spark SQL thrift server fails to pass settings to following query in the same 
> session
> -------------------------------------------------------------------------------------
>
>                 Key: SPARK-6540
>                 URL: https://issues.apache.org/jira/browse/SPARK-6540
>             Project: Spark
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 1.2.1
>            Reporter: Alex Liu
>
> e.g.
> By using beeline
> > set cassandra.username=user;
> > set cassandra.password = pass;
> > show databases;
> The "show databases" query fails due to missing cassandra.username and 
> cassandra.password configuration in hiveconf when create the metastore client.
> Same test passes in HiveServer2.
> This document 
> https://cwiki.apache.org/confluence/display/Hive/HiveServer2+Thrift+API 
> mentions how HiveServer2 maps the connection to session(explicit support for 
> sessions in the client API, e.g every RPC call references a session ID which 
> the server then maps to persistent session state.), but Spark Hive thrift 
> server fails to delivery the same function though it wraps HiveServer2 
> internally.
> Spark Hive thrift server contains a single hive context  which has a single 
> hiveConf shared by all sessions. Ideally we should isolate it per session.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to