You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Adam Budde (JIRA)" <ji...@apache.org> on 2016/12/23 20:32:58 UTC
[jira] [Comment Edited] (SPARK-18379) Make the parallelism of
parallelPartitionDiscovery configurable.
[ https://issues.apache.org/jira/browse/SPARK-18379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15773629#comment-15773629 ]
Adam Budde edited comment on SPARK-18379 at 12/23/16 8:32 PM:
--------------------------------------------------------------
Is there any reason this can't be pulled into a 2.1.x release? The hardcoded value introduced here in Spark 2.0.x is quite large and has caused a lot of performance regressions on our 1000+ core production clusters.
was (Author: budde):
Is there any reason this can't be pulled into a 2.1.x release? The hardcoded value introduced here in Spark 2.0.x is, to be frank, absurdly large and has caused a lot of performance regressions on our 1000+ core production clusters. This issue has had even worse for our smaller prototyping clusters, where "SELECT... LIMIT 1" queries now take several minutes to complete as a few hundred cores need to churn through 10000 partitions.
> Make the parallelism of parallelPartitionDiscovery configurable.
> -----------------------------------------------------------------
>
> Key: SPARK-18379
> URL: https://issues.apache.org/jira/browse/SPARK-18379
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 2.0.1
> Reporter: Genmao Yu
> Assignee: Genmao Yu
> Priority: Minor
> Fix For: 2.2.0
>
>
> The largest parallelism in PartitioningAwareFileIndex #listLeafFilesInParallel() is 10000 in hard code. We may need to make this number configurable. And in PR, I reduce it to 100.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org