[
https://issues.apache.org/jira/browse/HIVE-7387?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14058302#comment-14058302
]
Chengxiang Li commented on HIVE-7387:
-------------------------------------
Hive on spark can submit spark job in a separate JVM while
"hive.exec.submitviachild" configured as true, so that Hive only need to load
dependencies identified by HIVE-7371. But we still get this issue while Hive
submit spark job in same JVM with Hive CLI.
> Guava version conflict between hadoop and spark
> -----------------------------------------------
>
> Key: HIVE-7387
> URL: https://issues.apache.org/jira/browse/HIVE-7387
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Reporter: Chengxiang Li
>
> hadoop-hdfs and hadoop-comman have dependency on guava-11.0.2.jar, and spark
> dependent on guava-14.0.1.jar. guava-11.0.2 has API conflict with
> guava-14.0.1, as Hive CLI load both dependency into classpath currently,
> query failed on either spark engine or mr engine.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
