Mailing-List: contact hbase-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hbase-dev@hadoop.apache.org
Message-ID: <970111204.1238194730741.JavaMail.jira@brutus>
Date: Fri, 27 Mar 2009 15:58:50 -0700 (PDT)
From: "ryan rawson (JIRA)" <jira@apache.org>
To: hbase-dev@hadoop.apache.org
Subject: [jira] Updated: (HBASE-1293) hfile doesn't recycle decompressors
In-Reply-To: <912016106.1238194610637.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/HBASE-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

ryan rawson updated HBASE-1293:
-------------------------------

    Attachment: HBASE-1293.patch

> hfile doesn't recycle decompressors
> -----------------------------------
>
>                 Key: HBASE-1293
>                 URL: https://issues.apache.org/jira/browse/HBASE-1293
>             Project: Hadoop HBase
>          Issue Type: Bug
>    Affects Versions: 0.20.0
>         Environment: - all -
>            Reporter: ryan rawson
>            Assignee: ryan rawson
>             Fix For: 0.20.0
>
>         Attachments: HBASE-1293.patch
>
>
> The Compression codec stuff from hadoop has the concept of recycling compressors and decompressors - this is because a compression codec uses "direct buffers" which reside outside the JVM regular heap space.  There is a risk that under heavy concurrent load we could run out of that 'direct buffer' heap space in the JVM.
> HFile does not call algorithm.returnDecompressor and returnCompressor.  We should fix that.
> I found this bug via OOM crashes under jdk 1.7 - it appears to be partially due to the size of my cluster (200gb, 800 regions, 19 servers) and partially due to weaknesses in JVM 1.7.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.