[
https://issues.apache.org/jira/browse/HIVE-12951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15604832#comment-15604832
]
Richard Xin commented on HIVE-12951:
------------------------------------
Hi, we are facing issues while using hive-on-spark prewarm feature. (CDH 5.7.2,
HIve 1.1.0, spark 1.6.1 on a small cluster of 5 hosts)
here is what we did:
1. on CM, we changed
hive.prewarm.enabled to true,
hive.execution.engine to spark;
hive.prewarm.numcontainers to 5
spark.dynamicAllocation.minExecutors = 5;
spark.dynamicAllocation.enabled=true
2. restart hive
3. when we start a new hive session, the first query is still slow, subsequent
ones are much faster, It seems that the prewarm config didn't take effect.
Please advise what we have missed.
Thanks
> Reduce Spark executor prewarm timeout to 5s
> -------------------------------------------
>
> Key: HIVE-12951
> URL: https://issues.apache.org/jira/browse/HIVE-12951
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Affects Versions: 1.2.0
> Reporter: Xuefu Zhang
> Assignee: Xuefu Zhang
> Fix For: 2.1.0
>
> Attachments: HIVE-12951.patch, HIVE-12951.patch
>
>
> Currently it's set to 30s, which tends to be longer than needed. Reduce it to
> 5s, only considering jvm startup time. (Eventually, we may want to make this
> configurable.)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
