Mailing-List: contact hdfs-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hdfs-issues@hadoop.apache.org
Message-ID: <752932757.140471268074887511.JavaMail.jira@brutus.apache.org>
Date: Mon, 8 Mar 2010 19:01:27 +0000 (UTC)
From: "Todd Lipcon (JIRA)" <jira@apache.org>
To: hdfs-issues@hadoop.apache.org
Subject: [jira] Created: (HDFS-1029) Image corrupt with number of files = 1
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Image corrupt with number of files = 1
--------------------------------------

                 Key: HDFS-1029
                 URL: https://issues.apache.org/jira/browse/HDFS-1029
             Project: Hadoop HDFS
          Issue Type: Bug
          Components: name-node
    Affects Versions: 0.20.1
            Reporter: Todd Lipcon


Last week I recovered a corrupt namenode image that was completely sane except that the "number of files" in the header was set to 1, rather than the correct number (many million). The NN in question had been running for some time, so I believe the 2NN uploaded this broken image as a checkpoint. After this point, of course, no further checkpoints occurred, and the NN failed to load its image upon restart.

Not sure how this happens - my only thought is that we may need to add synchronization on the nsCount field in INodeDirectoryWithQuota, but that's a long shot.

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