hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wei-Chiu Chuang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-15697) memory leak in distcp run method
Date Mon, 27 Aug 2018 13:08:00 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-15697?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16593620#comment-16593620
] 

Wei-Chiu Chuang commented on HADOOP-15697:
------------------------------------------

Hi [~maheshk114] thanks for reporting the issue. Do you have a heap dump after the OOM? Without
a heapdump or a stacktrace it's hard to know what went wrong exactly.

One source of OOM that we are aware of is the file lists (distcp has to get a list of every
files) and it can grow to several gigabytes for a large directory. This known issue is addressed
by HADOOP-12502. But there could be other issues as well so we'll need more data to understand
better.

> memory leak in distcp run method
> --------------------------------
>
>                 Key: HADOOP-15697
>                 URL: https://issues.apache.org/jira/browse/HADOOP-15697
>             Project: Hadoop Common
>          Issue Type: Bug
>            Reporter: mahesh kumar behera
>            Priority: Major
>
> in distcp run method, execute is called but the job is not closed. This is causing OOM
error in Hive replication. I think in run method , shouldBlock should be set to true and execute
should be called within try-with-resources.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org


Mime
View raw message