Mailing-List: contact hdfs-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hdfs-issues@hadoop.apache.org
Message-ID: <2600067.181511272187850937.JavaMail.jira@thor>
Date: Sun, 25 Apr 2010 05:30:50 -0400 (EDT)
From: "dhruba borthakur (JIRA)" <jira@apache.org>
To: hdfs-issues@hadoop.apache.org
Subject: [jira] Commented: (HDFS-1108) ability to create a file whose newly
 allocated blocks are automatically persisted immediately
In-Reply-To: <5581135.169781272065570123.JavaMail.jira@thor>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HDFS-1108?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12860670#action_12860670 ] 

dhruba borthakur commented on HDFS-1108:
----------------------------------------

This is related to namenode HA.

suppose an application has created a file with one block and started writing data to that block. The writer has not yet written a full block worth of data to the file. Now, the NN fails over to the hot standby. The writer who was writing data should continue to write data to the file and should not see any interruption at all (assuming that the failover was done in a few seconds). For this use case, we need the ability to persist block allocations as soon as the block is allocated to a file.

The above could be alternatively achieved by making the DFSClient always issue a fsync for every new block allocation. This is not efficient because this translates to two RPCs for every new block allocation. Does this make sense?

> ability to create a file whose newly allocated blocks are automatically persisted immediately
> ---------------------------------------------------------------------------------------------
>
>                 Key: HDFS-1108
>                 URL: https://issues.apache.org/jira/browse/HDFS-1108
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: name-node
>            Reporter: dhruba borthakur
>            Assignee: dhruba borthakur
>
> The current HDFS design says that newly allocated blocks for a file are not persisted in the NN transaction log when the block is allocated. Instead, a hflush() or a close() on the file persists the blocks into the transaction log. It would be nice if we can immediately persist newly allocated blocks (as soon as they are allocated) for specific files.

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