hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yi Liu (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-8899) Erasure Coding: use threadpool for EC recovery tasks
Date Tue, 15 Sep 2015 00:48:46 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14744594#comment-14744594
] 

Yi Liu commented on HDFS-8899:
------------------------------

Thanks [~rakeshr] for the work and [~zhz] to ping me.

Overall looks good.  My comment is: 
Currently we have hard limit for replication/striped block reconstruction work on each datanode,
 the current maximum value is 4 by default if not configured.
So the default threadpool size in the patch is 20 which is too big and wasteful, also please
don't create a fix thread pool.  Maybe we can create a threadpool with {{corePoolSize}} is
2, and {{maximumPoolSize}} is 8?

> Erasure Coding: use threadpool for EC recovery tasks
> ----------------------------------------------------
>
>                 Key: HDFS-8899
>                 URL: https://issues.apache.org/jira/browse/HDFS-8899
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>            Reporter: Rakesh R
>            Assignee: Rakesh R
>         Attachments: HDFS-8899-HDFS-7285-01.patch, HDFS-8899-HDFS-7285-merge-00.patch
>
>
> The idea is to use threadpool for processing erasure coding recovery tasks at the datanode.
> {code}
> new Daemon(new ReconstructAndTransferBlock(recoveryInfo)).start();
> {code}



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

Mime
View raw message