[
https://issues.apache.org/jira/browse/SPARK-24764?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hyukjin Kwon resolved SPARK-24764.
----------------------------------
Resolution: Incomplete
> Add ServiceLoader implementation for SparkHadoopUtil
> ----------------------------------------------------
>
> Key: SPARK-24764
> URL: https://issues.apache.org/jira/browse/SPARK-24764
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 2.3.0, 2.3.1
> Reporter: Shruti Gumma
> Assignee: Shruti Gumma
> Priority: Major
> Labels: bulk-closed
>
> Currently SparkHadoopUtil creation is static and cannot be changed. This
> proposal is to move the creation of SparkHadoopUtil to a ServiceLoader, so
> that external cluster managers can create their own SparkHadoopUtil.
> In the case of Yarn, a specific YarnSparkHadoopUtil is used in the Yarn
> packages whereas in other places, SparkHadoopUtil is used. SparkHadoopUtil
> has been changed in v2.3.0 to work with Yarn, leaving external cluster
> managers with no configurable way of doing the same.
> The util classes such as SparkHadoopUtil should be made configurable, similar
> to how the ExternalClusterManager was made pluggable through a ServiceLoader.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
