[
https://issues.apache.org/jira/browse/PHOENIX-1128?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gabriel Reid updated PHOENIX-1128:
----------------------------------
Attachment: PHOENIX-1128.patch
Patch which reverts to using hadoop1 as the default profile (instead of
hadoop2), and corrects the dependencies for hadoop1 and hadoop2 for Pig.
The reason the hadoop1 profile needs to be the default is because of the
dependency on flume. Flume's poms also make use of a profile based on the
hadoop.profile property, and also activate hadoop1 based on the absence of the
hadoop.profile property. This means that if hadoop1 is provided as a property
to a build, it disables all profile handling in flume, and causes a broken
build in hadoop1 for us.
Another option would be to totally remove support for hadoop1.
> Fix build errors introduced by switch to hadoop2 as default profile
> -------------------------------------------------------------------
>
> Key: PHOENIX-1128
> URL: https://issues.apache.org/jira/browse/PHOENIX-1128
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 5.0.0, 4.1
> Reporter: Gabriel Reid
> Assignee: Gabriel Reid
> Attachments: PHOENIX-1128.patch
>
>
> PHOENIX-177 introduced some build issues (particularly for the Pig and Flume
> builds). This is largely due to the default build for 4.x being switched to
> hadoop2 instead of hadoop1.
> This ticket is to track the full fixing of the build since the introduction
> of PHOENIX-177.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
