[jira] [Commented] (HUDI-5430) Fix multi-writer handling w/ rollback blocks in MOR table (log record reader)

["sivabalan narayanan (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HUDI-5430?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17675769#comment-17675769 ] 

sivabalan narayanan commented on HUDI-5430:
-------------------------------------------

this is fixed with 5407 and 5408. There won't be any interleaved commits 

> Fix multi-writer handling w/ rollback blocks in MOR table (log record reader)
> -----------------------------------------------------------------------------
>
>                 Key: HUDI-5430
>                 URL: https://issues.apache.org/jira/browse/HUDI-5430
>             Project: Apache Hudi
>          Issue Type: Bug
>          Components: reader-core
>            Reporter: sivabalan narayanan
>            Assignee: sivabalan narayanan
>            Priority: Blocker
>             Fix For: 0.13.0
>
>
> Sample log blocks and commits
>  
> lb1_c1, lb2_c2, lb3_c3, lb4_c4, lb5_rb_c3, lb6_c5.
>  
> lb3 is expected to be considered invalid. but our current scan does not treat it as invalid. While parsing the rollback block, we just check for the previous log block for matching timestamp. since it does not match, we treat lb5_rb block as invalid and move on.
>  
> in case of multi-writer this is definitely feasible. so we should fix this.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)