Mailing-List: contact issues-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Date: Fri, 21 Apr 2017 21:26:04 +0000 (UTC)
From: =?utf-8?Q?Sergio_Pe=C3=B1a_=28JIRA=29?= <jira@apache.org>
To: issues@hive.apache.org
Message-ID: <JIRA.13047446.1488400006000.22599.1492809964277@Atlassian.JIRA>
In-Reply-To: <JIRA.13047446.1488400006000@Atlassian.JIRA>
References: <JIRA.13047446.1488400006000@Atlassian.JIRA> <JIRA.13047446.1488400006228@jira-lw-us.apache.org>
Subject: [jira] [Commented] (HIVE-16079) HS2: high memory pressure due to
 duplicate Properties objects
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Fri, 21 Apr 2017 21:26:09 -0000


    [ https://issues.apache.org/jira/browse/HIVE-16079?page=3Dcom.atlassian=
.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D1597=
9413#comment-15979413 ]=20

Sergio Pe=C3=B1a commented on HIVE-16079:
------------------------------------

The patch looks good [~misha@cloudera.com]
+1

I'll wait for the tests before commit it.

> HS2: high memory pressure due to duplicate Properties objects
> -------------------------------------------------------------
>
>                 Key: HIVE-16079
>                 URL: https://issues.apache.org/jira/browse/HIVE-16079
>             Project: Hive
>          Issue Type: Improvement
>          Components: HiveServer2
>            Reporter: Misha Dmitriev
>            Assignee: Misha Dmitriev
>         Attachments: HIVE-16079.01.patch, HIVE-16079.02.patch, HIVE-16079=
.03.patch, hs2-crash-2000p-500m-50q.txt
>
>
> I've created a Hive table with 2000 partitions, each backed by two files,=
 with one row in each file. When I execute some number of concurrent querie=
s against this table, e.g. as follows
> {code}
> for i in `seq 1 50`; do beeline -u jdbc:hive2://localhost:10000 -n admin =
-p admin -e "select count(i_f_1) from misha_table;" & done
> {code}
> it results in a big memory spike. With 20 queries I caused an OOM in a HS=
2 server with -Xmx200m and with 50 queries - in the one with -Xmx500m.
> I am attaching the results of jxray (www.jxray.com) analysis of a heap du=
mp that was generated in the 50queries/500m heap scenario. It suggests that=
 there are several opportunities to reduce memory pressure with not very in=
vasive changes to the code. One (duplicate strings) has been addressed in h=
ttps://issues.apache.org/jira/browse/HIVE-15882 In this ticket, I am going =
to address the fact that almost 20% of memory is used by instances of java.=
util.Properties. These objects are highly duplicate, since for each partiti=
on each concurrently running query creates its own copy of Partion, Partiti=
onDesc and Properties. Thus we have nearly 100,000 (50 queries * 2,000 part=
itions) Properties in memory. By interning/deduplicating these objects we m=
ay be able to save perhaps 15% of memory.
> Note, however, that if there are queries that mutate partitions, the corr=
esponding Properties would be mutated as well. Thus we cannot simply use a =
single "canonicalized" Properties object at all times for all Partition obj=
ects representing the same DB partition. Instead, I am going to introduce a=
 special CopyOnFirstWriteProperties class. Such an object initially interna=
lly references a canonicalized Properties object, and keeps doing so while =
only read methods are called. However, once any mutating method is called, =
the given CopyOnFirstWriteProperties copies the data into its own table fro=
m the canonicalized table, and uses it ever after.


--
This message was sent by Atlassian JIRA
(v6.3.15#6346)