[HUDI-6359] Spark offline compaction/clustering will never rollback when both requested and inflight states exist

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Fix Version/s: 1.1.0
• Component/s: clustering, compaction, table-service
• Labels: pull-request-available
• Assignee: Unassigned
• Reporter: kwang
• Created: 12/Jun/23 08:08
• Updated: 28/Mar/24 00:56

Description

If spark offline compaction execution fails, both a.compaction.requested and a.compaction.inflight states exist at the same time, which will cause the next  offline compaction never rolledback for compaction plan a.compaction.requested. Spark offline clustering is same as compaction.

Issue Links

• links to: GitHub Pull Request #8944