Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Date: Wed, 7 Jan 2015 15:25:34 +0000 (UTC)
From: "Cosmin Lehene (JIRA)" <jira@apache.org>
To: issues@hbase.apache.org
Message-ID: <JIRA.12502610.1301333897000.28715.1420644334836@Atlassian.JIRA>
In-Reply-To: <JIRA.12502610.1301333897000@Atlassian.JIRA>
References: <JIRA.12502610.1301333897000@Atlassian.JIRA>
 <JIRA.12502610.1301333897086@arcas>
Subject: [jira] [Commented] (HBASE-3706) On node crash (RS+DN), the
 replication of blocks can up latency
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HBASE-3706?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14267737#comment-14267737 ] 

Cosmin Lehene commented on HBASE-3706:
--------------------------------------

[~saint.ack@gmail.com] HDFS-1195 is stale... I'm assuming the same behavior still applies? Would it make sense to refresh this?

> On node crash (RS+DN), the replication of blocks can up latency
> ---------------------------------------------------------------
>
>                 Key: HBASE-3706
>                 URL: https://issues.apache.org/jira/browse/HBASE-3706
>             Project: HBase
>          Issue Type: Bug
>            Reporter: stack
>
> HDFS is kinda dumb regards rereplication in that if it notices missing replicas, its crass in the way it goes about making redress; there is no facility for limiting the rate of rereplication (HDFS-1195).  We've seen the case where hdfs is rereplicating at such a rate after DN crash, it fills the network hampering data serving.  This issue is about noting this prob. and helping out/keeping an eye on HDFS-1195.


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