[HUDI-4992] Spark Row-writing Bulk Insert produces incorrect Bloom Filter metadata

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 0.12.0
• Fix Version/s: 0.12.1
• Labels: pull-request-available
• Story Points: 2
• Epic Link: Hudi Spark Datasource
• Assignee: Alexey Kudinkin
• Reporter: Alexey Kudinkin
• Created: 06/Oct/22 23:13
• Updated: 18/Oct/22 20:13
• Resolved: 07/Oct/22 15:42

Description

Troubleshooting duplicates issue w/ Abhishek Modi from Notion, we've found that the min/max record key stats are being currently persisted incorrectly into Parquet metadata, leading to duplicate records being produced in their pipeline after initial bulk-insert.

Issue Links

• is a parent of: HUDI-5051 Add a functional regression test for Bloom Index followed on w/ Upserts
• links to: GitHub Pull Request #6883