Mailing-List: contact issues-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Date: Wed, 12 Jul 2017 07:59:02 +0000 (UTC)
From: "Chao Sun (JIRA)" <jira@apache.org>
To: issues@hive.apache.org
Message-ID: <JIRA.13084444.1499154499000.226006.1499846342441@Atlassian.JIRA>
In-Reply-To: <JIRA.13084444.1499154499000@Atlassian.JIRA>
References: <JIRA.13084444.1499154499000@Atlassian.JIRA> <JIRA.13084444.1499154499777@jira-lw-us.apache.org>
Subject: [jira] [Commented] (HIVE-17018) Small table is converted to map
 join even the total size of small tables exceeds the
 threshold(hive.auto.convert.join.noconditionaltask.size)
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Wed, 12 Jul 2017 07:59:49 -0000


    [ https://issues.apache.org/jira/browse/HIVE-17018?page=3Dcom.atlassian=
.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D1608=
3610#comment-16083610 ]=20

Chao Sun commented on HIVE-17018:
---------------------------------

[~kellyzly] Yes. I think we don't need to change the existing behavior. I'm=
 just suggesting that we might need a HoS specific config to replace {{hive=
.auto.convert.join.nonconditionaltask.size}}, so that it is less confusing.

> Small table is converted to map join even the total size of small tables =
exceeds the threshold(hive.auto.convert.join.noconditionaltask.size)
> -------------------------------------------------------------------------=
--------------------------------------------------------------------
>
>                 Key: HIVE-17018
>                 URL: https://issues.apache.org/jira/browse/HIVE-17018
>             Project: Hive
>          Issue Type: Bug
>            Reporter: liyunzhang_intel
>            Assignee: liyunzhang_intel
>         Attachments: HIVE-17018_data_init.q, HIVE-17018.q, t3.txt
>
>
>  we use "hive.auto.convert.join.noconditionaltask.size" as the threshold.=
 it means  the sum of size for n-1 of the tables/partitions for a n-way joi=
n is smaller than it, it will be converted to a map join. for example, A jo=
in B join C join D join E. Big table is A(100M), small tables are B(10M),C(=
10M),D(10M),E(10M).  If we set hive.auto.convert.join.noconditionaltask.siz=
e=3D20M. In current code, E,D,B will be converted to map join but C will no=
t be converted to map join. In my understanding, because hive.auto.convert.=
join.noconditionaltask.size can only contain E and D, so C and B should not=
 be converted to map join. =20
> Let's explain more why E can be converted to map join.
> in current code, [SparkMapJoinOptimizer#getConnectedMapJoinSize|https://g=
ithub.com/apache/hive/blob/master/ql/src/java/org/apache/hadoop/hive/ql/opt=
imizer/spark/SparkMapJoinOptimizer.java#L364] calculates all the mapjoins  =
in the parent path and child path. The search stops when encountering [Unio=
nOperator or ReduceOperator|https://github.com/apache/hive/blob/master/ql/s=
rc/java/org/apache/hadoop/hive/ql/optimizer/spark/SparkMapJoinOptimizer.jav=
a#L381]. Because C is not converted to map join because {{connectedMapJoinS=
ize + totalSize) > maxSize}} [see code|https://github.com/apache/hive/blob/=
master/ql/src/java/org/apache/hadoop/hive/ql/optimizer/spark/SparkMapJoinOp=
timizer.java#L330].The RS before the join of C remains. When calculating wh=
ether B will be converted to map join, {{getConnectedMapJoinSize}} returns =
0 as encountering [RS |https://github.com/apache/hive/blob/master/ql/src/ja=
va/org/apache/hadoop/hive/ql/optimizer/spark/SparkMapJoinOptimizer.java#409=
] and causes  {{connectedMapJoinSize + totalSize) < maxSize}} matches.
> [~xuefuz] or [~jxiang]: can you help see whether this is a bug or not  as=
 you are more familiar with SparkJoinOptimizer.


--
This message was sent by Atlassian JIRA
(v6.4.14#64029)