Started by timer2187208608f75307ccc19f0fb99dc8ddcce817240f66208608f75307ccc19f0fb99dc8ddcce817240f66refs/remotes/origin/main208608f75307ccc19f0fb99dc8ddcce817240f66208608f75307ccc19f0fb99dc8ddcce817240f66refs/remotes/origin/main208608f75307ccc19f0fb99dc8ddcce817240f66208608f75307ccc19f0fb99dc8ddcce817240f66refs/remotes/origin/mainhttps://gitbox.apache.org/repos/asf/lucene.git038619279testReportfalseJava: 64bit/hotspot/jdk-25.0.3 -XX:+UseCompressedOops -XX:+UseSerialGC#218732470364277901Lucene-nightly-main-Linux #21872187falsefalse2187238907SUCCESS1793370518482https://jenkins.thetaphi.de/view/Lucene/job/Lucene-nightly-main-Linux/2187/lucene/core/src/java/org/apache/lucene/index/DocumentsWriterFlushControl.javalucene/CHANGES.txtlucene/core/src/java/org/apache/lucene/index/FlushByRamOrCountsPolicy.javalucene/core/src/test/org/apache/lucene/index/TestIndexWriterDelete.java2813efa40a3c61228a1ec3ae4d230772ef8a8c331793355870000https://jenkins.thetaphi.de/user/noreplynoreplynoreply@github.comAvoid overflows in IndexWriter's delete buffer (#16867) The delete queue for IndexWriter is shared between all indexing threads, and holds a ByteBlockPool that tracks all terms being used for delete queries. While the individual indexing threads have their own maximum buffer size of 1945Mb to avoid overflowing any internal data structures, it is possible for this shared buffer to overflow if multiple threads are adding to it, and a particularly slow flush prevents it from being cleared. This adds a specific limit to the delete buffer (at the same size as the limits in DocumentsWriterPerThread), and also wires it into stall control so that other threads don't prevent the buffer being flushed once it hits the limit. 2026-10-30 10:24:30 +00002813efa40a3c61228a1ec3ae4d230772ef8a8c33Avoid overflows in IndexWriter's delete buffer (#16867)editlucene/core/src/java/org/apache/lucene/index/FlushByRamOrCountsPolicy.javaeditlucene/CHANGES.txteditlucene/core/src/java/org/apache/lucene/index/DocumentsWriterFlushControl.javaeditlucene/core/src/test/org/apache/lucene/index/TestIndexWriterDelete.javalucene/core/src/test/org/apache/lucene/search/TestTopFieldCollectorEarlyTermination.javalucene/core/src/java/org/apache/lucene/search/FieldValueHitQueue.javalucene/CHANGES.txtf6e64aa0db37d386d434d2dd3aac68665478407f1793356683000https://jenkins.thetaphi.de/user/noreplynoreplynoreply@github.comDisable sort-based early termination for indexes with document blocks (#16947) When an index defines a parent-child structure using IndexWriterConfig# setParentField, index sorting only looks at field values on parent documents. This means that sort-based early termination cannot use the value of its latest document as a cut-off, because it could be from a child document and later parent documents would be competitive. This commit disables sort-based early termination for indexes marked this way. Field-based pruning via competitive iterators continues to work as normal. 2026-10-30 10:38:03 +0000f6e64aa0db37d386d434d2dd3aac68665478407fDisable sort-based early termination for indexes with document blocks (#16947)editlucene/CHANGES.txteditlucene/core/src/test/org/apache/lucene/search/TestTopFieldCollectorEarlyTermination.javaeditlucene/core/src/java/org/apache/lucene/search/FieldValueHitQueue.javalucene/core/src/test/org/apache/lucene/search/TestSortOptimization.java46432a3f0db981f8d1ef5e6bdc142f0704b7c3211793362842000https://jenkins.thetaphi.de/user/noreplynoreplynoreply@github.comFix flaky TestSortOptimization.testDocSortOptimization (#16954) The hard-coded ceiling of 10 collected docs in this test can be breached if we randomly have more than 7 segments. Use numDocs instead, in line with other tests. 2026-10-30 12:20:42 +000046432a3f0db981f8d1ef5e6bdc142f0704b7c321Fix flaky TestSortOptimization.testDocSortOptimization (#16954)editlucene/core/src/test/org/apache/lucene/search/TestSortOptimization.javalucene/queries/src/test/org/apache/lucene/queries/TestCommonTermsQuery.java208608f75307ccc19f0fb99dc8ddcce817240f661793369113000https://jenkins.thetaphi.de/user/noreplynoreplynoreply@github.comFix flaky TestCommonTermsQuery test (#16955) Some similarities will give the same score to tail documents, and if RandomIndexWriter has re-ordered things behind the scenes then the expected doc ordering will break. Use a RandomIndexWriter that does no re-ordering. 2026-10-30 14:05:13 +0000208608f75307ccc19f0fb99dc8ddcce817240f66Fix flaky TestCommonTermsQuery test (#16955)editlucene/queries/src/test/org/apache/lucene/queries/TestCommonTermsQuery.javagithttps://jenkins.thetaphi.de/user/noreplynoreply