mikemccand opened a new issue, #14994: URL: https://github.com/apache/lucene/issues/14994
### Description Another spinoff when debugging an issue with `addIndexes`... ([first issue here](https://github.com/apache/lucene/issues/14993)): I noticed this odd `InfoStream` output): ``` DW 0 [2025-07-24T19:55:10.070573494Z; Lucene Merge Thread #36]: startFullFlush IW 0 [2025-07-24T19:55:10.070581863Z; Lucene Merge Thread #29]: now apply all deletes for all segments buffered updates bytesUsed=0 reader pool bytesUsed=0 BD 0 [2025-07-24T19:55:10.070663660Z; Lucene Merge Thread #29]: waitApply: no deletes to apply DW 0 [2025-07-24T19:55:10.070692476Z; Lucene Merge Thread #36]: Lucene Merge Thread #36 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.070776988Z; Lucene Merge Thread #35]: startFullFlush DW 0 [2025-07-24T19:55:10.070871338Z; Lucene Merge Thread #35]: Lucene Merge Thread #35 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.071129733Z; Lucene Merge Thread #34]: startFullFlush DW 0 [2025-07-24T19:55:10.071233683Z; Lucene Merge Thread #34]: Lucene Merge Thread #34 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.071301646Z; Lucene Merge Thread #33]: startFullFlush DW 0 [2025-07-24T19:55:10.071393444Z; Lucene Merge Thread #33]: Lucene Merge Thread #33 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.071456598Z; Lucene Merge Thread #32]: startFullFlush DW 0 [2025-07-24T19:55:10.071546108Z; Lucene Merge Thread #32]: Lucene Merge Thread #32 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.071612421Z; Lucene Merge Thread #30]: startFullFlush DW 0 [2025-07-24T19:55:10.071703621Z; Lucene Merge Thread #30]: Lucene Merge Thread #30 finishFullFlush success=true DW 0 [2025-07-24T19:55:10.071776826Z; Lucene Merge Thread #31]: startFullFlush DW 0 [2025-07-24T19:55:10.071911373Z; Lucene Merge Thread #31]: Lucene Merge Thread #31 finishFullFlush success=true ``` This is odd/smelly -- full flush is normally only done once, each time we get an NRT reader, commit/close, etc. I think [this line](https://github.com/apache/lucene/blob/branch_10_2/lucene/core/src/java/org/apache/lucene/index/IndexWriter.java#L3376) is the root cause -- that method (`addIndexesReaderMerge`) is the internal `IndexWriter` method called by each merge thread spawned by `addIndexes(CodecReader[])`. Why do we need to flush here, per merge thread, instead of [here](https://github.com/apache/lucene/blob/branch_10_2/lucene/core/src/java/org/apache/lucene/index/IndexWriter.java#L3167), once, up front, when `addIndexes` starts? Also, shouldn't `addIndexesReaderMerge` be `private`? It's the method each merge thread calls to actually do merging (during `addIndexes`)... ### Version and environment details _No response_ -- 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: issues-unsubscr...@lucene.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
