[
https://issues.apache.org/jira/browse/HUDI-2478?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17450759#comment-17450759
]
Prashant Wason commented on HUDI-2478:
--------------------------------------
The recovery today [is due to this
code|https://github.com/apache/hudi/blob/e0285800fb5609fcf27b523109e369a8d7d6b27d/hudi-client/hudi-client-common/src/main/java/org/apache/hudi/metadata/HoodieBackedTableMetadataWriter.java#L480:16]
writing another empty delete block into the log files.
The names of the bucket files are hardcoded per partition (files-0001,
files-0002, etc). So if we failed after creating a few of them, the next time
around the files will be appended to again. This has no affect due to the empty
delete block used to initialize the bucket log files.
So it works except for the rare case that the user changed the configured
bucket count in between the failed and successful operation.
> Handle failure mid-way during init buckets
> ------------------------------------------
>
> Key: HUDI-2478
> URL: https://issues.apache.org/jira/browse/HUDI-2478
> Project: Apache Hudi
> Issue Type: Sub-task
> Reporter: sivabalan narayanan
> Assignee: Vinoth Chandar
> Priority: Blocker
> Fix For: 0.10.0
>
>
> If process crashes mid-way while instantiating buckets, if tried again, it
> should work seamlessly.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
