[jira] [Commented] (ZOOKEEPER-1558) Leader should not snapshot uncommitted state

["Thawan Kooburat (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-1558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13782369#comment-13782369 ] 

Thawan Kooburat commented on ZOOKEEPER-1558:
--------------------------------------------

I think the patch need to be rebased in order to apply. 

With the recent issue of ZOOKEEPER-1697, I am wondering if we need to revisit the solution.  If I understand correctly, this patch move the snapshot step from pre-leader election to after synchronization and before broadcast phase. I believe we already transition to use syncLimit by the time we take this snapshot. I believe we should be taking snapshot while using initLimit  

In any case, we should add more comment about timing consideration into the patch as well.   

> Leader should not snapshot uncommitted state
> --------------------------------------------
>
>                 Key: ZOOKEEPER-1558
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1558
>             Project: ZooKeeper
>          Issue Type: Sub-task
>          Components: quorum
>            Reporter: Flavio Junqueira
>            Assignee: Flavio Junqueira
>            Priority: Blocker
>             Fix For: 3.4.6
>
>         Attachments: ZOOKEEPER-1558.patch, ZOOKEEPER-1558.patch, ZOOKEEPER-1558.patch, ZOOKEEPER-1558.patch
>
>
> Leader currently takes a snapshot when it calls loadData in the beginning of the lead() method. The loaded data, however, may contain uncommitted state.



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