Mailing-List: contact hama-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hama-dev@incubator.apache.org
Date: Thu, 15 Mar 2012 12:15:38 +0000 (UTC)
From: "Edward J. Yoon (Commented) (JIRA)" <jira@apache.org>
To: hama-dev@incubator.apache.org
Message-ID: 
 <1892711006.18736.1331813738299.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: 
 <1576816078.809.1328144753168.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Commented] (HAMA-499) Refactor clearZKNodes() in BSPMaster
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HAMA-499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13230097#comment-13230097 ] 

Edward J. Yoon commented on HAMA-499:
-------------------------------------

Sorry, I can't check right now. But, if I remember correctly, depth 2 (or 3) nodes won't deleted. So I wrote 3 nested for loop. And remained nodes in memory were really problematic. 
                
> Refactor clearZKNodes() in BSPMaster
> ------------------------------------
>
>                 Key: HAMA-499
>                 URL: https://issues.apache.org/jira/browse/HAMA-499
>             Project: Hama
>          Issue Type: Improvement
>          Components: bsp
>    Affects Versions: 0.4.0
>            Reporter: Edward J. Yoon
>              Labels: newbie
>             Fix For: 0.5.0
>
>         Attachments: HAMA-499.patch, HAMA-499_2.patch, move.sh
>
>
> As you know, Hama uses ZooKeepr for a barrier synchronization. User can use Hama ZK, or their own existing ZK cluster. To avoid some potential problems, BSPMaster calls clearZKNodes() to delete all remained sub-nodes of 'bspRoot' at initialization stage. That code is here and very ugly:
> {code}
>   public void clearZKNodes() {
>     try {
>       for (String node : zk.getChildren(bspRoot, this)) {
>         for (String subnode : zk.getChildren(bspRoot + "/" + node, this)) {
>           for (String subnode2 : zk.getChildren(bspRoot + "/" + node, this)) {
>             for (String subnode3 : zk.getChildren(bspRoot + "/" + node + "/" + subnode2, this)) {
>               zk.delete(bspRoot + "/" + node + "/" + subnode + "/" + subnode2 + "/" + subnode3, 0);
>             }
>             zk.delete(bspRoot + "/" + node + "/" + subnode + "/" + subnode2, 0);
>           }
>           zk.delete(bspRoot + "/" + node + "/" + subnode, 0);
>         }
>         zk.delete(bspRoot + "/" + node, 0);
>       }
>     } catch (KeeperException e) {
>     } catch (InterruptedException e) {
>     }
>   }
> {code}
> In this issue, we'll refactor this method. For example, you can delete recursively.
> P.S., must use '-1' to delete all version of each node like this:
> {code}
> zk.delete(node, -1);
> {code}

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira