You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by "Sagar Sumit (Jira)" <ji...@apache.org> on 2022/07/26 15:37:00 UTC

[jira] [Commented] (HUDI-4253) Fix filegroup and commit time for an index type when indexing with FILES partition

    [ https://issues.apache.org/jira/browse/HUDI-4253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17571507#comment-17571507 ] 

Sagar Sumit commented on HUDI-4253:
-----------------------------------

Fixed by https://github.com/apache/hudi/pull/5892/files

> Fix filegroup and commit time for an index type when indexing with FILES partition
> ----------------------------------------------------------------------------------
>
>                 Key: HUDI-4253
>                 URL: https://issues.apache.org/jira/browse/HUDI-4253
>             Project: Apache Hudi
>          Issue Type: Bug
>            Reporter: Sagar Sumit
>            Assignee: Sagar Sumit
>            Priority: Blocker
>             Fix For: 0.12.0
>
>
> This is a followup of the discussion in HUDI-4156.
> ```
> When both FILES and COLUMN_STATS are being initialized together, FILES partition takes the timestamp of latest completed commit in data table timeline, while COLUMN_STATS takes up new index commit time. and while enabling COLUMN_STATS partition, FSview does not return any file slices and so we end up setting file group count for COLUMN_STATS to 1.
> ```
> We need to ensure that once the metadata partition is initialized, then the subsequent files in the partition pick the corresponding commit time from the timeline.



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