[
https://issues.apache.org/jira/browse/PIO-49?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15879449#comment-15879449
]
ASF GitHub Bot commented on PIO-49:
-----------------------------------
Github user pferrel commented on the issue:
https://github.com/apache/incubator-predictionio/pull/336
So you are using the old PIO plugins feature that basically puts any files
in `plugins` onto the SparkSubmit --files param? The plugins are, in effect,
jars outside the pio assembly to include when running a job?
This might work since we are not linking in classes from more than one
version of ES but how do you move a jar that isn't built, how do the jars get
built? Seems like some part of the build process has to change even if it's to
create these alternative jars?
> Add support for Elasticsearch 5.x
> ---------------------------------
>
> Key: PIO-49
> URL: https://issues.apache.org/jira/browse/PIO-49
> Project: PredictionIO
> Issue Type: Improvement
> Reporter: Shinsuke Sugaya
>
> We work on meta/event storage support for Elasticsearch 5.x.
> Although Elasticsearch 2.x does not allow dots in field names,
> Elasticsearch 5.x supports it. So, it's better to upgrade to ES 5.x release.
> Since ES 5.x provides Java Rest API client, we replaced
> Transport communication with HTTP one. Therefore, our fix
> uses HTTP(9200 port) only.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
