[
https://issues.apache.org/jira/browse/HIVE-9251?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rui Li updated HIVE-9251:
-------------------------
Attachment: HIVE-9251.3-spark.patch
Addressed RB comments and updated golden files. Some notes about the reducer
count changes:
* Most queries changed from 3 to 2. We're using total cores to set # reducer
here. But previously we counted driver as an executor so we have 1 more
executor as we really do. Actually neither count is correct because in fact we
have 4 cores (2 executors each with 2 cores), however we can't get cores per
executor info.
* Some queries changed from 3 to 1. That's because {{hive.exec.reducers.max}}
is set to 1 but we previously didn't respect it.
* Some queries need deterministic results and that's tracked by HIVE-9290.
> SetSparkReducerParallelism is likely to set too small number of reducers
> [Spark Branch]
> ---------------------------------------------------------------------------------------
>
> Key: HIVE-9251
> URL: https://issues.apache.org/jira/browse/HIVE-9251
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Reporter: Rui Li
> Assignee: Rui Li
> Attachments: HIVE-9251.1-spark.patch, HIVE-9251.2-spark.patch,
> HIVE-9251.3-spark.patch
>
>
> This may hurt performance or even lead to task failures. For example, spark's
> netty-based shuffle limits the max frame size to be 2G.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
