Mailing-List: contact hdfs-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hdfs-issues@hadoop.apache.org
Date: Wed, 24 Aug 2011 17:52:30 +0000 (UTC)
From: "Todd Lipcon (JIRA)" <jira@apache.org>
To: hdfs-issues@hadoop.apache.org
Message-ID: 
 <1001740525.10050.1314208350789.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Created] (HDFS-2283) Already pending replication targets
 are incorrectly chosen for another replication of the same block
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Already pending replication targets are incorrectly chosen for another replication of the same block
----------------------------------------------------------------------------------------------------

                 Key: HDFS-2283
                 URL: https://issues.apache.org/jira/browse/HDFS-2283
             Project: Hadoop HDFS
          Issue Type: Bug
          Components: name-node
    Affects Versions: 0.23.0
            Reporter: Todd Lipcon


In looking at tests for HDFS-1172, I came across this issue: we currently only track the _number_ of pending replications per block, but not the targets themselves. Thus, if the replication count of a block is raised while replicas are already pending, the second set of replication targets may overlap with the first. This causes a ReplicaAlreadyExistsException for the second replication attempt, and the block is left under-replicated until the pending replication times out.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira