lasdf1234 opened a new pull request, #14675: URL: https://github.com/apache/gravitino/pull/14675
### What changes were proposed in this pull request? For partition-aware maintenance strategies on partitioned tables, stop seeding trigger/score evaluation with table-level statistics. Evaluation now uses partition statistics plus shared table metadata and policy rules only. Remove the table-context-only short-circuit in `shouldTriggerForPartitionTable`. ### Why are the changes needed? `builtin-iceberg-update-stats` refreshes partition statistics for partitioned tables and does not rewrite table-level keys. Seeding partition evaluation with table-level statistics let stale values such as `custom-data-file-mse` drive `submit-strategy-jobs` recommendations. Fixes #14674 ### Does this PR introduce _any_ user-facing change? Yes. Partition-aware strategies on partitioned tables decide trigger/score from partition statistics (and metadata/rules), not from table-level statistic values. Non-partitioned tables are unchanged. No configuration keys are added. ### How was this patch tested? Updated and extended `TestCompactionStrategyHandler`, including a regression where a large stale table-level MSE must not trigger when every partition is below the threshold. Ran: ``` ./gradlew :maintenance:optimizer:test --tests 'org.apache.gravitino.maintenance.optimizer.recommender.handler.compaction.TestCompactionStrategyHandler' -PskipITs ``` Made with [Cursor](https://cursor.com) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
