You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by "zhihao song (Jira)" <ji...@apache.org> on 2023/03/31 02:57:00 UTC
[jira] [Commented] (HUDI-6007) When using the MOR table with flink, hudi savepoint is invalid
[ https://issues.apache.org/jira/browse/HUDI-6007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17707090#comment-17707090 ]
zhihao song commented on HUDI-6007:
-----------------------------------
I think savepoint is not completely invalid when using the MOR table with Flink, since mor table also have basefiles when finish compaction.No saving the logfiles just could potentially lead to a consistency issues.Can you edit the description? Thanks~
> When using the MOR table with flink, hudi savepoint is invalid
> --------------------------------------------------------------
>
> Key: HUDI-6007
> URL: https://issues.apache.org/jira/browse/HUDI-6007
> Project: Apache Hudi
> Issue Type: Bug
> Reporter: zouxxyy
> Priority: Major
>
> Currently hudi's savepoint only saves the base file, and filter the files in it when clean.
> But when using the MOR table with flink, hudi's savepoint is actually invalid, because there is no base file.
> file:
> {code:java}
> .4e164104-da91-4c39-89b8-a81890cf2c8c_20230330235205548.log.1_0-1-0
> .4e164104-da91-4c39-89b8-a81890cf2c8c_20230330235205548.log.2_0-1-0
> .hoodie_partition_metadata {code}
> 20230330235205548.savepoint:
> {code:java}
> {
> "savepointedBy": "",
> "savepointedAt": 1680191759153,
> "comments": "",
> "partitionMetadata": {
> "dt=2021-12-09/hh=10": {
> "partitionPath": "dt=2021-12-09/hh=10",
> "savepointDataFile": []
> }
> },
> "version": {
> "int": 1
> }
> } {code}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)