You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kylin.apache.org by "liyang (JIRA)" <ji...@apache.org> on 2015/09/16 11:58:45 UTC
[jira] [Commented] (KYLIN-742) Route unsupported queries to
SparkSQL
[ https://issues.apache.org/jira/browse/KYLIN-742?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14747240#comment-14747240 ]
liyang commented on KYLIN-742:
------------------------------
Thanks Murali for the patch! I can see it works but requires some polishing.
- The config properties need better naming. The prefix {{kylin.jdbc.hive2}} is a bit confusing cause people may think it's how Kylin read hive data for cube building. But actually it's for routing. How about {{kylin.route.hive.xx}}?
- Need a global config to turn this feature on/off. {{kylin.route.hive.enabled}}? Be off by default.
- The default JDBC URL could be {{jdbc:hive2://}}? According to hive doc [1], this means embedded server, seems more generic than {{jdbc:hive2://sandbox:10000}}. Can it work?
- Be selective about what exception to route to hive. As to design doc [2], only SqlValidatorException and ParseException worth retrying. Other exceptions should fail out loud, not be hidden by retrying hive.
- Please avoid lengthy code in catch block. Instead extract and encapsulate Hive query code in a new class.
[1] https://cwiki.apache.org/confluence/display/Hive/HiveServer2+Clients#HiveServer2Clients-JDBC
[2] https://issues.apache.org/jira/secure/attachment/12744202/Design%20Documentation.pdf
> Route unsupported queries to SparkSQL
> -------------------------------------
>
> Key: KYLIN-742
> URL: https://issues.apache.org/jira/browse/KYLIN-742
> Project: Kylin
> Issue Type: New Feature
> Components: Query Engine, SparkSQL
> Reporter: Luke Han
> Assignee: liyang
> Fix For: Backlog
>
> Attachments: 1.x-HBase1.x-KYLIN-742-update.patch, Design Documentation.pdf
>
>
> When Kylin can't serve coming SQL, it will be better to route it to enabled SQL on Hadoop like SparkSQL and execute there.
> Then get result and return to client through Kylin server.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)