You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Dan Burkert (JIRA)" <ji...@apache.org> on 2017/04/03 22:45:42 UTC
[jira] [Commented] (KUDU-1454) Spark and MR jobs running without
scan locality
[ https://issues.apache.org/jira/browse/KUDU-1454?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15954274#comment-15954274 ]
Dan Burkert commented on KUDU-1454:
-----------------------------------
Now that KUDU-1715 has landed, is fixing this as simple as setting the replica selection to {{CLOSEST_REPLICA}}, or do we need to be using fault tolerant scans as well?
> Spark and MR jobs running without scan locality
> -----------------------------------------------
>
> Key: KUDU-1454
> URL: https://issues.apache.org/jira/browse/KUDU-1454
> Project: Kudu
> Issue Type: Bug
> Components: client, perf, spark
> Affects Versions: 0.8.0
> Reporter: Todd Lipcon
> Priority: Critical
>
> Spark (and according to [~danburkert] MR also now) add all of the locations of a tablet as split locations. This makes sense except that the Java client currently always scans the leader replica. So in many cases we schedule a task which is "local" to a follower, and then it ends up having to do a remote scan.
> This makes Spark queries take about twice as long on tables with replicas compared to unreplicated tables, and I think is a regression on the MR side.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)