Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Message-ID: <28894814.172601292565961064.JavaMail.jira@thor>
Date: Fri, 17 Dec 2010 01:06:01 -0500 (EST)
From: "stack (JIRA)" <jira@apache.org>
To: issues@hbase.apache.org
Subject: [jira] Commented: (HBASE-3366) WALObservers should be notified
 before the lock
In-Reply-To: <2329869.150051292463541629.JavaMail.jira@thor>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HBASE-3366?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12972364#action_12972364 ] 

stack commented on HBASE-3366:
------------------------------

You are saying that the newly rolled file, the one that is created inside the synchronized block, that it got some edits and these edits were not "observed" because notification of roll happened after the new file was created? 

If so, +1 on your patch.

> WALObservers should be notified before the lock
> -----------------------------------------------
>
>                 Key: HBASE-3366
>                 URL: https://issues.apache.org/jira/browse/HBASE-3366
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Jean-Daniel Cryans
>            Assignee: Jean-Daniel Cryans
>             Fix For: 0.90.0
>
>         Attachments: HBASE-3366.patch
>
>
> TestReplication failed on the last 0.90 build because the log of the machine that was killed was rolled and got a few edits in before it was put in a ZK queue.
> This fix is pretty easy, the notification should be sent before the lock is acquired instead of after that. It shouldn't have been like that since the beginning but I guess I missed it when I narrowed the lock.

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