beam-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Amit Sela (JIRA)" <>
Subject [jira] [Commented] (BEAM-375) HadoopIO and runners-spark conflict with hadoop.version
Date Sat, 25 Jun 2016 07:43:41 GMT


Amit Sela commented on BEAM-375:

But do we want Beam Hadoop components to support multiple Hadoop versions ? with a default
build like Spark does ?

Correct me if I'm wrong here, but going with the latest is not the best choice anyway (unless
there is a constraint) because AFAIK Hadoop is not something users upgrade as fast as say,
Spark/Flink versions.


> HadoopIO and runners-spark conflict with hadoop.version
> -------------------------------------------------------
>                 Key: BEAM-375
>                 URL:
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-java-extensions
>            Reporter: Pei He
>            Assignee: Pei He
> HadoopIO currently uses 2.7.0 and runners-spark uses 2.2.0 for hadoop-client, hadoop-common.
> From [~amitsela]
> "Spark can be built against different hadoop versions, but the release in maven central
is a 2.2.0 build (latest). ''
> For HadoopIO, I don't know why 2.7.0 is picked at the beginning. I can check if it will
work with 2.2.0.
> I am creating this issue, since I think it there is a general question.
> In principle, HadoopIO and other sdks Sources should work with any runners. But, when
one set of runners require version A, but the other set of runners require version B, we will
need a general solution for it.

This message was sent by Atlassian JIRA

View raw message