Throws and exception for radial search when mapping is for on-disk mo…

…de (opensearch-project#2055) Signed-off-by: Tejas Shah <[email protected]>
heemin32 · Sep 7, 2024 · cbc6343 · cbc6343
1 parent 7bf52ac
commit cbc6343
Show file tree

Hide file tree

Showing 8 changed files with 103 additions and 4 deletions.
diff --git a/src/main/java/org/opensearch/knn/index/mapper/KNNMappingConfig.java b/src/main/java/org/opensearch/knn/index/mapper/KNNMappingConfig.java
@@ -6,6 +6,7 @@
 package org.opensearch.knn.index.mapper;
 
 import org.opensearch.knn.index.engine.KNNMethodContext;
+import org.opensearch.knn.index.engine.qframe.QuantizationConfig;
 
 import java.util.Optional;
 
@@ -48,6 +49,14 @@ default CompressionLevel getCompressionLevel() {
         return CompressionLevel.NOT_CONFIGURED;
     }
 
+    /**
+     * Returns quantization config
+     * @return
+     */
+    default QuantizationConfig getQuantizationConfig() {
+        return QuantizationConfig.EMPTY;
+    }
+
     /**
      *
      * @return the dimension of the index; for model based indices, it will be null

diff --git a/src/main/java/org/opensearch/knn/index/mapper/KNNVectorFieldMapper.java b/src/main/java/org/opensearch/knn/index/mapper/KNNVectorFieldMapper.java
@@ -241,7 +241,8 @@ public KNNVectorFieldMapper build(BuilderContext context) {
                     hasDocValues.get(),
                     modelDao,
                     indexCreatedVersion,
-                    originalParameters
+                    originalParameters,
+                    knnMethodConfigContext
                 );
             }
 

diff --git a/src/main/java/org/opensearch/knn/index/mapper/MethodFieldMapper.java b/src/main/java/org/opensearch/knn/index/mapper/MethodFieldMapper.java
@@ -51,6 +51,12 @@ public static MethodFieldMapper createFieldMapper(
         boolean hasDocValues,
         OriginalMappingParameters originalMappingParameters
     ) {
+
+        KNNMethodContext knnMethodContext = originalMappingParameters.getResolvedKnnMethodContext();
+        QuantizationConfig quantizationConfig = knnMethodContext.getKnnEngine()
+            .getKNNLibraryIndexingContext(knnMethodContext, knnMethodConfigContext)
+            .getQuantizationConfig();
+
         final KNNVectorFieldType mappedFieldType = new KNNVectorFieldType(
             fullname,
             metaValue,
@@ -75,6 +81,11 @@ public Mode getMode() {
                 public CompressionLevel getCompressionLevel() {
                     return knnMethodConfigContext.getCompressionLevel();
                 }
+
+                @Override
+                public QuantizationConfig getQuantizationConfig() {
+                    return quantizationConfig;
+                }
             }
         );
         return new MethodFieldMapper(

diff --git a/src/main/java/org/opensearch/knn/index/mapper/ModelFieldMapper.java b/src/main/java/org/opensearch/knn/index/mapper/ModelFieldMapper.java
@@ -56,9 +56,17 @@ public static ModelFieldMapper createFieldMapper(
         boolean hasDocValues,
         ModelDao modelDao,
         Version indexCreatedVersion,
-        OriginalMappingParameters originalMappingParameters
+        OriginalMappingParameters originalMappingParameters,
+        KNNMethodConfigContext knnMethodConfigContext
     ) {
 
+        final KNNMethodContext knnMethodContext = originalMappingParameters.getKnnMethodContext();
+        final QuantizationConfig quantizationConfig = knnMethodContext == null
+            ? QuantizationConfig.EMPTY
+            : knnMethodContext.getKnnEngine()
+                .getKNNLibraryIndexingContext(knnMethodContext, knnMethodConfigContext)
+                .getQuantizationConfig();
+
         final KNNVectorFieldType mappedFieldType = new KNNVectorFieldType(fullname, metaValue, vectorDataType, new KNNMappingConfig() {
             private Integer dimension = null;
             private Mode mode = null;
@@ -94,6 +102,11 @@ public CompressionLevel getCompressionLevel() {
                 return compressionLevel;
             }
 
+            @Override
+            public QuantizationConfig getQuantizationConfig() {
+                return quantizationConfig;
+            }
+
             // ModelMetadata relies on cluster state which may not be available during field mapper creation. Thus,
             // we lazily initialize it.
             private void initFromModelMetadata() {

diff --git a/src/main/java/org/opensearch/knn/index/query/KNNQueryBuilder.java b/src/main/java/org/opensearch/knn/index/query/KNNQueryBuilder.java
@@ -25,6 +25,7 @@
 import org.opensearch.index.query.QueryShardContext;
 import org.opensearch.knn.index.engine.KNNMethodConfigContext;
 import org.opensearch.knn.index.engine.model.QueryContext;
+import org.opensearch.knn.index.engine.qframe.QuantizationConfig;
 import org.opensearch.knn.index.mapper.KNNMappingConfig;
 import org.opensearch.knn.index.mapper.KNNVectorFieldType;
 import org.opensearch.knn.index.query.parser.RescoreParser;
@@ -451,6 +452,10 @@ protected Query doToQuery(QueryShardContext context) {
             if (vectorDataType == VectorDataType.BINARY) {
                 throw new UnsupportedOperationException(String.format(Locale.ROOT, "Binary data type does not support radial search"));
             }
+
+            if (knnMappingConfig.getQuantizationConfig() != QuantizationConfig.EMPTY) {
+                throw new UnsupportedOperationException("Radial search is not supported for indices which have quantization enabled");
+            }
         }
 
         // Currently, k-NN supports distance and score types radial search

diff --git a/src/main/java/org/opensearch/knn/index/query/nativelib/NativeEngineKnnVectorQuery.java b/src/main/java/org/opensearch/knn/index/query/nativelib/NativeEngineKnnVectorQuery.java
@@ -7,6 +7,7 @@
 
 import lombok.Getter;
 import lombok.RequiredArgsConstructor;
+import lombok.extern.log4j.Log4j2;
 import org.apache.lucene.index.IndexReader;
 import org.apache.lucene.index.LeafReaderContext;
 import org.apache.lucene.search.IndexSearcher;
@@ -18,6 +19,7 @@
 import org.apache.lucene.search.Weight;
 import org.apache.lucene.util.BitSet;
 import org.apache.lucene.util.Bits;
+import org.opensearch.common.StopWatch;
 import org.opensearch.knn.index.query.KNNQuery;
 import org.opensearch.knn.index.query.KNNWeight;
 import org.opensearch.knn.index.query.ResultUtil;
@@ -39,6 +41,7 @@
  * for k-NN query if required. This is done by overriding rewrite method to execute ANN on each leaf
  * {@link KNNQuery} does not give the ability to post process segment results.
  */
+@Log4j2
 @Getter
 @RequiredArgsConstructor
 public class NativeEngineKnnVectorQuery extends Query {
@@ -60,7 +63,11 @@ public Weight createWeight(IndexSearcher indexSearcher, ScoreMode scoreMode, flo
             int firstPassK = rescoreContext.getFirstPassK(finalK);
             perLeafResults = doSearch(indexSearcher, leafReaderContexts, knnWeight, firstPassK);
             ResultUtil.reduceToTopK(perLeafResults, firstPassK);
+
+            StopWatch stopWatch = new StopWatch().start();
             perLeafResults = doRescore(indexSearcher, leafReaderContexts, knnWeight, perLeafResults, finalK);
+            long rescoreTime = stopWatch.stop().totalTime().millis();
+            log.debug("Rescoring results took {} ms. oversampled k:{}, segments:{}", rescoreTime, firstPassK, leafReaderContexts.size());
         }
         ResultUtil.reduceToTopK(perLeafResults, finalK);
         TopDocs[] topDocs = new TopDocs[perLeafResults.size()];

diff --git a/src/test/java/org/opensearch/knn/index/mapper/KNNVectorFieldMapperTests.java b/src/test/java/org/opensearch/knn/index/mapper/KNNVectorFieldMapperTests.java
@@ -1130,6 +1130,12 @@ public void testModelFieldMapperParseCreateField_validInput_thenDifferentFieldTy
             MockedStatic<KNNVectorFieldMapperUtil> utilMockedStatic = Mockito.mockStatic(KNNVectorFieldMapperUtil.class);
             MockedStatic<ModelUtil> modelUtilMockedStatic = Mockito.mockStatic(ModelUtil.class)
         ) {
+            KNNMethodConfigContext knnMethodConfigContext = KNNMethodConfigContext.builder()
+                .vectorDataType(VectorDataType.FLOAT)
+                .versionCreated(CURRENT)
+                .dimension(TEST_DIMENSION)
+                .build();
+
             for (VectorDataType dataType : VectorDataType.values()) {
                 log.info("Vector Data Type is : {}", dataType);
                 SpaceType spaceType = VectorDataType.BINARY == dataType ? SpaceType.DEFAULT_BINARY : SpaceType.INNER_PRODUCT;
@@ -1173,7 +1179,8 @@ public void testModelFieldMapperParseCreateField_validInput_thenDifferentFieldTy
                     false,
                     modelDao,
                     CURRENT,
-                    originalMappingParameters
+                    originalMappingParameters,
+                    knnMethodConfigContext
                 );
 
                 modelFieldMapper.parseCreateField(parseContext);
@@ -1214,7 +1221,8 @@ public void testModelFieldMapperParseCreateField_validInput_thenDifferentFieldTy
                     false,
                     modelDao,
                     CURRENT,
-                    originalMappingParameters
+                    originalMappingParameters,
+                    knnMethodConfigContext
                 );
 
                 modelFieldMapper.parseCreateField(parseContext);

diff --git a/src/test/java/org/opensearch/knn/index/query/KNNQueryBuilderTests.java b/src/test/java/org/opensearch/knn/index/query/KNNQueryBuilderTests.java
@@ -29,7 +29,10 @@
 import org.opensearch.index.query.QueryShardContext;
 import org.opensearch.index.query.TermQueryBuilder;
 import org.opensearch.knn.KNNTestCase;
+import org.opensearch.knn.index.engine.qframe.QuantizationConfig;
+import org.opensearch.knn.index.mapper.KNNMappingConfig;
 import org.opensearch.knn.index.mapper.KNNVectorFieldType;
+import org.opensearch.knn.index.mapper.Mode;
 import org.opensearch.knn.index.query.rescore.RescoreContext;
 import org.opensearch.knn.index.util.KNNClusterUtil;
 import org.opensearch.knn.index.engine.KNNMethodContext;
@@ -41,6 +44,7 @@
 import org.opensearch.knn.indices.ModelDao;
 import org.opensearch.knn.indices.ModelMetadata;
 import org.opensearch.knn.indices.ModelState;
+import org.opensearch.knn.quantization.enums.ScalarQuantizationType;
 
 import java.io.IOException;
 import java.util.Arrays;
@@ -440,6 +444,47 @@ public void testDoToQuery_whenRadialSearchOnBinaryIndex_thenException() {
         assertTrue(e.getMessage().contains("Binary data type does not support radial search"));
     }
 
+    public void testDoToQuery_whenRadialSearchOnDiskMode_thenException() {
+        float[] queryVector = { 1.0f };
+        KNNQueryBuilder knnQueryBuilder = KNNQueryBuilder.builder()
+            .fieldName(FIELD_NAME)
+            .vector(queryVector)
+            .maxDistance(MAX_DISTANCE)
+            .build();
+        Index dummyIndex = new Index("dummy", "dummy");
+        QueryShardContext mockQueryShardContext = mock(QueryShardContext.class);
+        KNNVectorFieldType mockKNNVectorField = mock(KNNVectorFieldType.class);
+        when(mockQueryShardContext.index()).thenReturn(dummyIndex);
+        when(mockKNNVectorField.getVectorDataType()).thenReturn(VectorDataType.FLOAT);
+        when(mockQueryShardContext.fieldMapper(anyString())).thenReturn(mockKNNVectorField);
+        MethodComponentContext methodComponentContext = new MethodComponentContext(
+            org.opensearch.knn.common.KNNConstants.METHOD_HNSW,
+            ImmutableMap.of()
+        );
+        KNNMethodContext knnMethodContext = new KNNMethodContext(KNNEngine.FAISS, SpaceType.L2, methodComponentContext);
+        when(mockKNNVectorField.getKnnMappingConfig()).thenReturn(new KNNMappingConfig() {
+            @Override
+            public Optional<KNNMethodContext> getKnnMethodContext() {
+                return Optional.of(knnMethodContext);
+            }
+
+            @Override
+            public int getDimension() {
+                return 1;
+            }
+
+            public Mode getMode() {
+                return Mode.ON_DISK;
+            }
+
+            public QuantizationConfig getQuantizationConfig() {
+                return QuantizationConfig.builder().quantizationType(ScalarQuantizationType.ONE_BIT).build();
+            }
+        });
+        Exception e = expectThrows(UnsupportedOperationException.class, () -> knnQueryBuilder.doToQuery(mockQueryShardContext));
+        assertEquals("Radial search is not supported for indices which have quantization enabled", e.getMessage());
+    }
+
     public void testDoToQuery_KnnQueryWithFilter_Lucene() throws Exception {
         // Given
         float[] queryVector = { 1.0f, 2.0f, 3.0f, 4.0f };