You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@pig.apache.org by "Koji Noguchi (JIRA)" <ji...@apache.org> on 2015/09/16 23:11:48 UTC

[jira] [Updated] (PIG-3251) Bzip2TextInputFormat requires double the memory of maximum record size

     [ https://issues.apache.org/jira/browse/PIG-3251?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Koji Noguchi updated PIG-3251:
------------------------------
    Attachment: pig-3251-trunk-v06.patch

Updated the patch (pig-3251-trunk-v06.patch) to
* Only to use Hadoop's TextInputFormat for 0.23/2.X.
* Make it configurable and turning it *on* as default.  (need a good config name)
* Moving the update of this fake codec with extension "bz" to PigServer.java.  (probably a wrong place but not sure where to put.)
* Updating TestBZip so that it would run the test twice for 0.23/2.X with option turned on and off.

Then I realized that for the issue I previously reported on this jira, testBlockHeaderEndingWithCR and testBlockHeaderEndingAtSplitNotByteAligned failing, only the former is fixed in  MAPREDUCE-5656 but not the latter.  
Looking at testBlockHeaderEndingAtSplitNotByteAligned.



> Bzip2TextInputFormat requires double the memory of maximum record size
> ----------------------------------------------------------------------
>
>                 Key: PIG-3251
>                 URL: https://issues.apache.org/jira/browse/PIG-3251
>             Project: Pig
>          Issue Type: Improvement
>            Reporter: Koji Noguchi
>            Assignee: Koji Noguchi
>            Priority: Minor
>             Fix For: 0.16.0
>
>         Attachments: pig-3251-trunk-v01.patch, pig-3251-trunk-v02.patch, pig-3251-trunk-v03.patch, pig-3251-trunk-v04.patch, pig-3251-trunk-v05.patch, pig-3251-trunk-v06.patch
>
>
> While looking at user's OOM heap dump, noticed that pig's Bzip2TextInputFormat consumes memory at both
> Bzip2TextInputFormat.buffer (ByteArrayOutputStream) 
> and actual Text that is returned as line.
> For example, when having one record with 160MBytes, buffer was 268MBytes and Text was 160MBytes.  
> We can probably eliminate one of them.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)