[jira] [Comment Edited] (ZOOKEEPER-1639) zk.getZKDatabase().deserializeSnapshot adds new system znodes instead of replacing existing ones

["Flavio Junqueira (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-1639?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789337#comment-13789337 ] 

Flavio Junqueira edited comment on ZOOKEEPER-1639 at 10/8/13 4:15 PM:
----------------------------------------------------------------------

I'm missing some context here. What kind of incorrect behavior does this bug induce? 


was (Author: fpj):
I'm missing some context here. What kind of incorrect behavior does this bug induces? 

> zk.getZKDatabase().deserializeSnapshot adds new system znodes instead of replacing existing ones
> ------------------------------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-1639
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1639
>             Project: ZooKeeper
>          Issue Type: Bug
>    Affects Versions: 3.4.5
>            Reporter: Alexander Shraer
>
> Before the call to zk.getZKDatabase().deserializeSnapshot in Learner.java, 
> zk.getZKDatabase().getDataTree().getNode("/zookeeper") == zk.getZKDatabase().getDataTree().procDataNode, which means that this is the same znode, as it should be. 
> However, after this call, they are not equal. The node actually being used in client operations is zk.getZKDatabase().getDataTree().getNode("/zookeeper"), but the other old node procDataNode is still there and not replaced (in fact it is a final field).



--
This message was sent by Atlassian JIRA
(v6.1#6144)