lucene_snapshot: Update to new Lucene 10.3 postings format (#128240)

There are a few different things going on in this PR, all of which are required to get the lucene_snspahot branch building again, but the most substantial is the update to the new Lucene 10.3 postings format, Lucene103PostingsFormat.

The Lucene90BlockTreeTermsWriter class is used in the implementation of the 10.1 postings codec in Lucene. With the new 10.3 postings format that class is no longer needed, so it has been moved to a test-only location, in order to support backward compatibility testing.

In Elasticsearch we were using Lucene90BlockTreeTermsWriter (from Lucene) directly, through our copy of the Lucene 9.0 postings format, namely ES812PostingsFormat. So if you look at ES812PostingsFormat , you should that the imports should now show that we're using our own copy.

Additionally, changes are required because of the removal of deprecated methods in IOContext, as well as the override of hints for direct IO.

Co-authored-by: Simon Cooper <[email protected]>

Author: ChrisHegarty

benchmarks/src/main/java/org/elasticsearch/benchmark/index/codec/tsdb/TSDBDocValuesMergeBenchmark.java

@@ -26,7 +26,7 @@
 import org.apache.lucene.util.BytesRef;
 import org.elasticsearch.cluster.metadata.DataStream;
 import org.elasticsearch.common.logging.LogConfigurator;
-import org.elasticsearch.index.codec.Elasticsearch900Lucene101Codec;
+import org.elasticsearch.index.codec.Elasticsearch92Lucene103Codec;
 import org.elasticsearch.index.codec.tsdb.es819.ES819TSDBDocValuesFormat;
 import org.openjdk.jmh.annotations.Benchmark;
 import org.openjdk.jmh.annotations.BenchmarkMode;
@@ -63,7 +63,6 @@
 public class TSDBDocValuesMergeBenchmark {
 
     static {
-        // For Elasticsearch900Lucene101Codec:
         LogConfigurator.loadLog4jPlugins();
         LogConfigurator.configureESLogging();
         LogConfigurator.setNodeName("test");
@@ -259,7 +258,7 @@ private static IndexWriterConfig createIndexWriterConfig(boolean optimizedMergeE
         config.setLeafSorter(DataStream.TIMESERIES_LEAF_READERS_SORTER);
         config.setMergePolicy(new LogByteSizeMergePolicy());
         var docValuesFormat = new ES819TSDBDocValuesFormat(4096, optimizedMergeEnabled);
-        config.setCodec(new Elasticsearch900Lucene101Codec() {
+        config.setCodec(new Elasticsearch92Lucene103Codec() {
 
             @Override
             public DocValuesFormat getDocValuesFormatForField(String field) {

server/src/internalClusterTest/java/org/elasticsearch/search/simple/SimpleSearchIT.java

@@ -261,20 +261,22 @@ public void testSimpleTerminateAfterCount() throws Exception {
         ensureGreen();
         refresh();
 
-        for (int i = 1; i < max; i++) {
+        // query all but one doc to avoid optimizations that may rewrite to a MatchAllDocs, which simplifies assertions
+        final int queryMax = max - 1;
+        for (int i = 1; i < queryMax; i++) {
             final int finalI = i;
             assertResponse(
-                prepareSearch("test").setQuery(QueryBuilders.rangeQuery("field").gte(1).lte(max)).setTerminateAfter(i),
+                prepareSearch("test").setQuery(QueryBuilders.rangeQuery("field").gte(1).lte(queryMax)).setTerminateAfter(i),
                 response -> {
                     assertHitCount(response, finalI);
                     assertTrue(response.isTerminatedEarly());
                 }
             );
         }
         assertResponse(
-            prepareSearch("test").setQuery(QueryBuilders.rangeQuery("field").gte(1).lte(max)).setTerminateAfter(2 * max),
+            prepareSearch("test").setQuery(QueryBuilders.rangeQuery("field").gte(1).lte(queryMax)).setTerminateAfter(2 * max),
             response -> {
-                assertHitCount(response, max);
+                assertHitCount(response, queryMax);
                 assertFalse(response.isTerminatedEarly());
             }
         );

server/src/main/java/module-info.java

@@ -461,7 +461,8 @@
             org.elasticsearch.index.codec.Elasticsearch814Codec,
             org.elasticsearch.index.codec.Elasticsearch816Codec,
             org.elasticsearch.index.codec.Elasticsearch900Codec,
-            org.elasticsearch.index.codec.Elasticsearch900Lucene101Codec;
+            org.elasticsearch.index.codec.Elasticsearch900Lucene101Codec,
+            org.elasticsearch.index.codec.Elasticsearch92Lucene103Codec;
 
     provides org.apache.logging.log4j.core.util.ContextDataProvider with org.elasticsearch.common.logging.DynamicContextDataProvider;
 

server/src/main/java/org/elasticsearch/action/admin/indices/diskusage/IndexDiskUsageAnalyzer.java

@@ -10,6 +10,7 @@
 package org.elasticsearch.action.admin.indices.diskusage;
 
 import org.apache.logging.log4j.Logger;
+import org.apache.lucene.backward_codecs.lucene101.Lucene101PostingsFormat;
 import org.apache.lucene.backward_codecs.lucene50.Lucene50PostingsFormat;
 import org.apache.lucene.backward_codecs.lucene84.Lucene84PostingsFormat;
 import org.apache.lucene.backward_codecs.lucene90.Lucene90PostingsFormat;
@@ -22,7 +23,7 @@
 import org.apache.lucene.codecs.PointsReader;
 import org.apache.lucene.codecs.StoredFieldsReader;
 import org.apache.lucene.codecs.TermVectorsReader;
-import org.apache.lucene.codecs.lucene101.Lucene101PostingsFormat;
+import org.apache.lucene.codecs.lucene103.Lucene103PostingsFormat;
 import org.apache.lucene.index.BinaryDocValues;
 import org.apache.lucene.index.ByteVectorValues;
 import org.apache.lucene.index.DirectoryReader;
@@ -318,6 +319,9 @@ private static void readProximity(Terms terms, PostingsEnum postings) throws IOE
     private static BlockTermState getBlockTermState(TermsEnum termsEnum, BytesRef term) throws IOException {
         if (term != null && termsEnum.seekExact(term)) {
             final TermState termState = termsEnum.termState();
+            if (termState instanceof final Lucene103PostingsFormat.IntBlockTermState blockTermState) {
+                return new BlockTermState(blockTermState.docStartFP, blockTermState.posStartFP, blockTermState.payStartFP);
+            }
             if (termState instanceof final Lucene101PostingsFormat.IntBlockTermState blockTermState) {
                 return new BlockTermState(blockTermState.docStartFP, blockTermState.posStartFP, blockTermState.payStartFP);
             }

server/src/main/java/org/elasticsearch/common/lucene/Lucene.java

@@ -93,7 +93,7 @@
 
 public class Lucene {
 
-    public static final String LATEST_CODEC = "Lucene101";
+    public static final String LATEST_CODEC = "Lucene103";
 
     public static final String SOFT_DELETES_FIELD = "__soft_deletes";
 

server/src/main/java/org/elasticsearch/index/codec/CodecService.java

@@ -12,7 +12,7 @@
 import org.apache.lucene.codecs.Codec;
 import org.apache.lucene.codecs.FieldInfosFormat;
 import org.apache.lucene.codecs.FilterCodec;
-import org.apache.lucene.codecs.lucene101.Lucene101Codec;
+import org.apache.lucene.codecs.lucene103.Lucene103Codec;
 import org.elasticsearch.common.util.BigArrays;
 import org.elasticsearch.common.util.FeatureFlag;
 import org.elasticsearch.core.Nullable;
@@ -46,7 +46,7 @@ public class CodecService implements CodecProvider {
     public CodecService(@Nullable MapperService mapperService, BigArrays bigArrays) {
         final var codecs = new HashMap<String, Codec>();
 
-        Codec legacyBestSpeedCodec = new LegacyPerFieldMapperCodec(Lucene101Codec.Mode.BEST_SPEED, mapperService, bigArrays);
+        Codec legacyBestSpeedCodec = new LegacyPerFieldMapperCodec(Lucene103Codec.Mode.BEST_SPEED, mapperService, bigArrays);
         if (ZSTD_STORED_FIELDS_FEATURE_FLAG) {
             codecs.put(DEFAULT_CODEC, new PerFieldMapperCodec(Zstd814StoredFieldsFormat.Mode.BEST_SPEED, mapperService, bigArrays));
         } else {
@@ -58,7 +58,7 @@ public CodecService(@Nullable MapperService mapperService, BigArrays bigArrays)
             BEST_COMPRESSION_CODEC,
             new PerFieldMapperCodec(Zstd814StoredFieldsFormat.Mode.BEST_COMPRESSION, mapperService, bigArrays)
         );
-        Codec legacyBestCompressionCodec = new LegacyPerFieldMapperCodec(Lucene101Codec.Mode.BEST_COMPRESSION, mapperService, bigArrays);
+        Codec legacyBestCompressionCodec = new LegacyPerFieldMapperCodec(Lucene103Codec.Mode.BEST_COMPRESSION, mapperService, bigArrays);
         codecs.put(LEGACY_BEST_COMPRESSION_CODEC, legacyBestCompressionCodec);
 
         codecs.put(LUCENE_DEFAULT_CODEC, Codec.getDefault());

server/src/main/java/org/elasticsearch/index/codec/Elasticsearch900Lucene101Codec.java

@@ -9,12 +9,12 @@
 
 package org.elasticsearch.index.codec;
 
+import org.apache.lucene.backward_codecs.lucene101.Lucene101Codec;
+import org.apache.lucene.backward_codecs.lucene101.Lucene101PostingsFormat;
 import org.apache.lucene.codecs.DocValuesFormat;
 import org.apache.lucene.codecs.KnnVectorsFormat;
 import org.apache.lucene.codecs.PostingsFormat;
 import org.apache.lucene.codecs.StoredFieldsFormat;
-import org.apache.lucene.codecs.lucene101.Lucene101Codec;
-import org.apache.lucene.codecs.lucene101.Lucene101PostingsFormat;
 import org.apache.lucene.codecs.lucene90.Lucene90DocValuesFormat;
 import org.apache.lucene.codecs.lucene99.Lucene99HnswVectorsFormat;
 import org.apache.lucene.codecs.perfield.PerFieldKnnVectorsFormat;

server/src/main/java/org/elasticsearch/index/codec/Elasticsearch92Lucene103Codec.java

@@ -0,0 +1,133 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the "Elastic License
+ * 2.0", the "GNU Affero General Public License v3.0 only", and the "Server Side
+ * Public License v 1"; you may not use this file except in compliance with, at
+ * your election, the "Elastic License 2.0", the "GNU Affero General Public
+ * License v3.0 only", or the "Server Side Public License, v 1".
+ */
+
+package org.elasticsearch.index.codec;
+
+import org.apache.lucene.codecs.DocValuesFormat;
+import org.apache.lucene.codecs.KnnVectorsFormat;
+import org.apache.lucene.codecs.PostingsFormat;
+import org.apache.lucene.codecs.StoredFieldsFormat;
+import org.apache.lucene.codecs.lucene103.Lucene103Codec;
+import org.apache.lucene.codecs.lucene103.Lucene103PostingsFormat;
+import org.apache.lucene.codecs.lucene90.Lucene90DocValuesFormat;
+import org.apache.lucene.codecs.lucene99.Lucene99HnswVectorsFormat;
+import org.apache.lucene.codecs.perfield.PerFieldKnnVectorsFormat;
+import org.apache.lucene.codecs.perfield.PerFieldPostingsFormat;
+import org.elasticsearch.index.codec.perfield.XPerFieldDocValuesFormat;
+import org.elasticsearch.index.codec.zstd.Zstd814StoredFieldsFormat;
+
+/**
+ * Elasticsearch codec as of 9.2 relying on Lucene 10.3. This extends the Lucene 10.3 codec to compressed
+ * stored fields with ZSTD instead of LZ4/DEFLATE. See {@link Zstd814StoredFieldsFormat}.
+ */
+public class Elasticsearch92Lucene103Codec extends CodecService.DeduplicateFieldInfosCodec {
+
+    static final PostingsFormat DEFAULT_POSTINGS_FORMAT = new Lucene103PostingsFormat();
+
+    private final StoredFieldsFormat storedFieldsFormat;
+
+    private final PostingsFormat defaultPostingsFormat;
+    private final PostingsFormat postingsFormat = new PerFieldPostingsFormat() {
+        @Override
+        public PostingsFormat getPostingsFormatForField(String field) {
+            return Elasticsearch92Lucene103Codec.this.getPostingsFormatForField(field);
+        }
+    };
+
+    private final DocValuesFormat defaultDVFormat;
+    private final DocValuesFormat docValuesFormat = new XPerFieldDocValuesFormat() {
+        @Override
+        public DocValuesFormat getDocValuesFormatForField(String field) {
+            return Elasticsearch92Lucene103Codec.this.getDocValuesFormatForField(field);
+        }
+    };
+
+    private final KnnVectorsFormat defaultKnnVectorsFormat;
+    private final KnnVectorsFormat knnVectorsFormat = new PerFieldKnnVectorsFormat() {
+        @Override
+        public KnnVectorsFormat getKnnVectorsFormatForField(String field) {
+            return Elasticsearch92Lucene103Codec.this.getKnnVectorsFormatForField(field);
+        }
+    };
+
+    /** Public no-arg constructor, needed for SPI loading at read-time. */
+    public Elasticsearch92Lucene103Codec() {
+        this(Zstd814StoredFieldsFormat.Mode.BEST_SPEED);
+    }
+
+    /**
+     * Constructor. Takes a {@link Zstd814StoredFieldsFormat.Mode} that describes whether to optimize for retrieval speed at the expense of
+     * worse space-efficiency or vice-versa.
+     */
+    public Elasticsearch92Lucene103Codec(Zstd814StoredFieldsFormat.Mode mode) {
+        super("Elasticsearch92Lucene103", new Lucene103Codec());
+        this.storedFieldsFormat = mode.getFormat();
+        this.defaultPostingsFormat = DEFAULT_POSTINGS_FORMAT;
+        this.defaultDVFormat = new Lucene90DocValuesFormat();
+        this.defaultKnnVectorsFormat = new Lucene99HnswVectorsFormat();
+    }
+
+    @Override
+    public StoredFieldsFormat storedFieldsFormat() {
+        return storedFieldsFormat;
+    }
+
+    @Override
+    public final PostingsFormat postingsFormat() {
+        return postingsFormat;
+    }
+
+    @Override
+    public final DocValuesFormat docValuesFormat() {
+        return docValuesFormat;
+    }
+
+    @Override
+    public final KnnVectorsFormat knnVectorsFormat() {
+        return knnVectorsFormat;
+    }
+
+    /**
+     * Returns the postings format that should be used for writing new segments of <code>field</code>.
+     *
+     * <p>The default implementation always returns "Lucene912".
+     *
+     * <p><b>WARNING:</b> if you subclass, you are responsible for index backwards compatibility:
+     * future version of Lucene are only guaranteed to be able to read the default implementation,
+     */
+    public PostingsFormat getPostingsFormatForField(String field) {
+        return defaultPostingsFormat;
+    }
+
+    /**
+     * Returns the docvalues format that should be used for writing new segments of <code>field</code>
+     * .
+     *
+     * <p>The default implementation always returns "Lucene912".
+     *
+     * <p><b>WARNING:</b> if you subclass, you are responsible for index backwards compatibility:
+     * future version of Lucene are only guaranteed to be able to read the default implementation.
+     */
+    public DocValuesFormat getDocValuesFormatForField(String field) {
+        return defaultDVFormat;
+    }
+
+    /**
+     * Returns the vectors format that should be used for writing new segments of <code>field</code>
+     *
+     * <p>The default implementation always returns "Lucene912".
+     *
+     * <p><b>WARNING:</b> if you subclass, you are responsible for index backwards compatibility:
+     * future version of Lucene are only guaranteed to be able to read the default implementation.
+     */
+    public KnnVectorsFormat getKnnVectorsFormatForField(String field) {
+        return defaultKnnVectorsFormat;
+    }
+
+}

server/src/main/java/org/elasticsearch/index/codec/LegacyPerFieldMapperCodec.java

@@ -13,7 +13,7 @@
 import org.apache.lucene.codecs.DocValuesFormat;
 import org.apache.lucene.codecs.KnnVectorsFormat;
 import org.apache.lucene.codecs.PostingsFormat;
-import org.apache.lucene.codecs.lucene101.Lucene101Codec;
+import org.apache.lucene.codecs.lucene103.Lucene103Codec;
 import org.elasticsearch.common.lucene.Lucene;
 import org.elasticsearch.common.util.BigArrays;
 import org.elasticsearch.index.mapper.MapperService;
@@ -22,11 +22,11 @@
  * Legacy version of {@link PerFieldMapperCodec}. This codec is preserved to give an escape hatch in case we encounter issues with new
  * changes in {@link PerFieldMapperCodec}.
  */
-public final class LegacyPerFieldMapperCodec extends Lucene101Codec {
+public final class LegacyPerFieldMapperCodec extends Lucene103Codec {
 
     private final PerFieldFormatSupplier formatSupplier;
 
-    public LegacyPerFieldMapperCodec(Lucene101Codec.Mode compressionMode, MapperService mapperService, BigArrays bigArrays) {
+    public LegacyPerFieldMapperCodec(Lucene103Codec.Mode compressionMode, MapperService mapperService, BigArrays bigArrays) {
         super(compressionMode);
         this.formatSupplier = new PerFieldFormatSupplier(mapperService, bigArrays);
         // If the below assertion fails, it is a sign that Lucene released a new codec. You must create a copy of the current Elasticsearch

server/src/main/java/org/elasticsearch/index/codec/PerFieldFormatSupplier.java

@@ -12,7 +12,6 @@
 import org.apache.lucene.codecs.DocValuesFormat;
 import org.apache.lucene.codecs.KnnVectorsFormat;
 import org.apache.lucene.codecs.PostingsFormat;
-import org.apache.lucene.codecs.lucene101.Lucene101PostingsFormat;
 import org.apache.lucene.codecs.lucene90.Lucene90DocValuesFormat;
 import org.apache.lucene.codecs.lucene99.Lucene99HnswVectorsFormat;
 import org.elasticsearch.common.util.BigArrays;
@@ -34,13 +33,12 @@
  * vectors.
  */
 public class PerFieldFormatSupplier {
-    public static final FeatureFlag USE_LUCENE101_POSTINGS_FORMAT = new FeatureFlag("use_lucene101_postings_format");
+    public static final FeatureFlag USE_DEFAULT_LUCENE_POSTINGS_FORMAT = new FeatureFlag("use_default_lucene_postings_format");
 
     private static final DocValuesFormat docValuesFormat = new Lucene90DocValuesFormat();
     private static final KnnVectorsFormat knnVectorsFormat = new Lucene99HnswVectorsFormat();
     private static final ES819TSDBDocValuesFormat tsdbDocValuesFormat = new ES819TSDBDocValuesFormat();
     private static final ES812PostingsFormat es812PostingsFormat = new ES812PostingsFormat();
-    private static final Lucene101PostingsFormat lucene101PostingsFormat = new Lucene101PostingsFormat();
     private static final PostingsFormat completionPostingsFormat = PostingsFormat.forName("Completion101");
 
     private final ES87BloomFilterPostingsFormat bloomFilterPostingsFormat;
@@ -53,10 +51,10 @@ public PerFieldFormatSupplier(MapperService mapperService, BigArrays bigArrays)
         this.bloomFilterPostingsFormat = new ES87BloomFilterPostingsFormat(bigArrays, this::internalGetPostingsFormatForField);
 
         if (mapperService != null
-            && USE_LUCENE101_POSTINGS_FORMAT.isEnabled()
-            && mapperService.getIndexSettings().getIndexVersionCreated().onOrAfter(IndexVersions.USE_LUCENE101_POSTINGS_FORMAT)
+            && USE_DEFAULT_LUCENE_POSTINGS_FORMAT.isEnabled()
+            && mapperService.getIndexSettings().getIndexVersionCreated().onOrAfter(IndexVersions.UPGRADE_TO_LUCENE_10_3_0)
             && mapperService.getIndexSettings().getMode() == IndexMode.STANDARD) {
-            defaultPostingsFormat = lucene101PostingsFormat;
+            defaultPostingsFormat = Elasticsearch92Lucene103Codec.DEFAULT_POSTINGS_FORMAT;
         } else {
             // our own posting format using PFOR
             defaultPostingsFormat = es812PostingsFormat;