[
https://issues.apache.org/jira/browse/PHOENIX-2568?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15085650#comment-15085650
]
Josh Mahonin commented on PHOENIX-2568:
---------------------------------------
Yup, it actually works the same as before on a stock spark install, these
issues came up when trying to use the client-spark JAR on the HDP platform.
They (and probably CDH) do some extra things with Hive and Hadoop that exercise
code paths which involve thrift, jackson and some of the SPI stuff.
This will probably get superseded by PHOENIX-2535 but it's a quick and easy fix
for now.
> phoenix-spark: client-spark JAR conflicting with some Spark distributions
> -------------------------------------------------------------------------
>
> Key: PHOENIX-2568
> URL: https://issues.apache.org/jira/browse/PHOENIX-2568
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.7.0
> Reporter: Josh Mahonin
> Assignee: Josh Mahonin
> Fix For: 4.7.0
>
> Attachments: PHOENIX-2568.patch
>
>
> The new client-spark JAR is still shipping with org.codehaus.jackson, as well
> as the SPI aggregation. This causes some distributions of Spark (e.g., HDP)
> to have classpath errors when using spark-shell.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
