RandomSamplingFacetsCollector.createManager samples the hits twice when a search has more than one slice: the reduce reads each slice's already-sampled getMatchingDocs() as its original hits, so the sample is far too small and every amortized facet count is biased low (regression from #13542, 10.0.0 onward)

Finding LFV-765Area: facet / sampling (RandomSamplingFacetsCollector.createManager reduce, FacetsCollectorManager.reduceMatchingDocs)Priority: mediumImpact: silent wrong resultsNovelty: new

Finding page for LFV-765

AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.

What it shows

You count facets on a sample of a large result set with RandomSamplingFacetsCollector.createManager(sampleSize, seed) on an IndexSearcher that has an executor, and expect what a one-slice search gives: about sampleSize sampled hits, taken at the rate sampleSize / totalHits that getSamplingRate() reports, so that amortizeFacetCounts estimates the real counts. Instead, with two or more slices the reduce fills the reduced collector's original hits from FacetsCollectorManager.reduceMatchingDocs, which calls each slice collector's getMatchingDocs(), the accessor RandomSamplingFacetsCollector overrides to return that slice's sample; the reduced collector then samples those samples again. The sample shrinks to about sampleSize^2 x numSlices / totalHits and every amortized count is low by the per-slice ratio, with no error: 2 segments of 1,024 matching docs with sampleSize 128 and one slice per segment give 16 sampled hits and an amortized count of 256 instead of 128 and 2,048, and default slicing on 2 segments of 262,144 docs with sampleSize 4,096 gives 64 sampled hits and 8,192 instead of 4,096 and 524,288.

The test shows that with two slices (one per segment), RandomSamplingFacetsCollector.createManager samples the hits twice: the reduce samples again the per-slice samples it reads through the overridden getMatchingDocs().

Expected vs actual

Expected: 128 sampled hits out of 2,048 (the reported rate 0.0625), as a one-slice search gives. Actual (-ea and -da): 16.

Files

Run it

# Lucene checkout at the tested commit, built with: ./gradlew --no-daemon assemble   (JDK 25)
LUCENE=/path/to/lucene; V=12.0.0-SNAPSHOT
CP=$LUCENE/lucene/core/build/libs/lucene-core-$V.jar
CP=$CP:$LUCENE/lucene/codecs/build/libs/lucene-codecs-$V.jar
CP=$CP:$LUCENE/lucene/test-framework/build/libs/lucene-test-framework-$V.jar
CP=$CP:$LUCENE/lucene/facet/build/libs/lucene-facet-$V.jar
CP=$CP:randomizedtesting-runner-2.9.2.jar:junit-4.13.2.jar:hamcrest-3.0.jar   # test-framework dependencies (Maven Central)
CP=$CP:junit-platform-console-standalone-6.1.3.jar   # JUnit Platform console launcher (Maven Central)

javac -d classes -cp "$CP" $(find src -name '*.java')

# run 1: assertions on, seed DEADBEEF
java -ea -esa --add-modules jdk.incubator.vector --enable-native-access=ALL-UNNAMED -Dtests.seed=DEADBEEF \
     -cp "classes:$CP" org.junit.platform.console.ConsoleLauncher \
     execute --select-class org.apache.lucene.facet.TestSamplingManagerTwoSlices

The module(s) facet also need their third-party runtime jars; ./gradlew -p lucene/<module> dependencies lists them.

Gradle alternative: copy the test into the matching module's src/test/ tree of the checkout and run ./gradlew -p lucene/<module> test --tests <class> -Ptests.seed=<seed> (add -Ptests.asserts=false for a run without assertions).

Feedback

Open an issue in lucene-ai-bug-hunt. Please do not open issues or pull requests in apache/lucene about this finding unless a Lucene committer agrees it should be filed.

Snapshot of 2026-11-03. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.