[jira] [Commented] (ZOOKEEPER-2855) Rebooting a Joined Node Failed Due The Joined Node Previously Failed to Update Its Configuration Correctly

["Alexander Shraer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-2855?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16099109#comment-16099109 ] 

Alexander Shraer commented on ZOOKEEPER-2855:
---------------------------------------------

Thanks Jeffrey for finding this issue. 

Looking at the code and the steps you wrote down, it seems that we're not doing the right thing when processing the packetsCommitted:

for(Long zxid: packetsCommitted) {
   fzk.commit(zxid);
}

This isn't doing any reconfig related processing, which should be similar to what's done when a COMMIT_AND_ACTIVATE is received.




> Rebooting a Joined Node Failed Due The Joined Node Previously Failed to Update Its Configuration Correctly
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-2855
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2855
>             Project: ZooKeeper
>          Issue Type: Bug
>          Components: leaderElection, quorum, server
>    Affects Versions: 3.5.3
>            Reporter: Jeffrey F. Lukman
>         Attachments: ZK-2855.pdf
>
>
> We are testing our distributed system model checking (DMCK)
> by directing our DMCK to reproduce the ZooKepeer-2172 bug in the ZooKeeper v3.5.3.
> After some exploration, our DMCK found that the ZooKeeper-2172 still linger in the reported fixed version, ZooKeeper v.3.5.3.
> Here we attached the complete bug scenario to reproduce the bug.
> We have communicated this bug to [~shralex] and he has confirmed that this bug exists.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)