[jira] [Commented] (BOOKKEEPER-581) Ledger recovery doesn't work correctly when recovery adds force changing ensembles.

["Ivan Kelly (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/BOOKKEEPER-581?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13677947#comment-13677947 ] 

Ivan Kelly commented on BOOKKEEPER-581:
---------------------------------------

Committed to branch-4.2 as r1490578. Some changes to imports had to be manually merged. Final patch uploaded.
                
> Ledger recovery doesn't work correctly when recovery adds force changing ensembles.
> -----------------------------------------------------------------------------------
>
>                 Key: BOOKKEEPER-581
>                 URL: https://issues.apache.org/jira/browse/BOOKKEEPER-581
>             Project: Bookkeeper
>          Issue Type: Bug
>          Components: bookkeeper-client
>    Affects Versions: 4.2.0, 4.2.1
>            Reporter: Sijie Guo
>            Assignee: Sijie Guo
>            Priority: Critical
>             Fix For: 4.2.2, 4.3.0
>
>         Attachments: BOOKKEEPER-581.branch4-2.diff, BOOKKEEPER-581.patch
>
>
> currently read and write use same ledger metadata during ledger recovery, which is bad. since write will cause changing ensemble which modified ensembles map to introducing brand new bookies. those brand new bookies would mislead following recovery reads, cause recovery is proceeded in a wrong way.
> E.g.
> 3 bookies, quorums size 2. A, B, C.
> read 0 from A, B.
> recovery add 0: A, B becomes slow. D, E are brought into the ensemble to replace A, B.
> so following recovery read would be proceed in ensemble (D, E, C), then we would lost all the entries added in A and B.
> this issue is similar as BOOKKEEPER-355.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira