[
https://issues.apache.org/jira/browse/PIG-5283?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16116449#comment-16116449
]
Rohini Palaniswamy commented on PIG-5283:
-----------------------------------------
bq. My only question is that if we should only write those properties that are
required for a PigSplit instead of writing the full jobConf (6-700 entries) for
optimization.
I would suggest trimming down and also see if it is possible to serialize
only once. You are serializing the config with each split which is not good.
That is a lot of overhead and will impact performance. Had run into performance
issues and OOMs with Tez on huge configs and serializing configs multiple times
and had to trim down.
> Configuration is not passed to SparkPigSplits on the backend
> ------------------------------------------------------------
>
> Key: PIG-5283
> URL: https://issues.apache.org/jira/browse/PIG-5283
> Project: Pig
> Issue Type: Bug
> Components: spark
> Reporter: Adam Szita
> Assignee: Adam Szita
> Attachments: PIG-5283.0.patch
>
>
> When a Hadoop ObjectWritable is created during a Spark job, the instantiated
> PigSplit (wrapped into a SparkPigSplit) is given an empty Configuration
> instance.
> This happens
> [here|https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/SerializableWritable.scala#L44]
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
