Avoid doing I/O when fetching min and max for keyword fields (#92026)

Whenever sorting on a date, numeric or keyword field (as primary sort), the can_match phase retrieves min and max for the field and sorts the shards (asc or desc depending on the sort order) so that they are going to be queried following that order. This allows incremental results to be exposed in that same order when using async search, as well as optimizations built on top of such behaviour (#51852).

For fields with points we call `getMinPackedValue` and `getMaxPackedValue`, while for keyword fields we call `Terms#getMin` and `Terms#getMax`. Elasticsearch uses `FilterTerms` implementations to cancel queries as well as to track field usage. Such filter implementations should delegate their `getMin` and `getMax` calls to the wrapped `Terms` instance, which will leverage info from the block tree that caches min and max, otherwise they are always going to be retrieved from the index, which does I/O and slows the can_match phase down.

docs/changelog/92026.yaml

@@ -0,0 +1,5 @@
+pr: 92026
+summary: Avoid doing I/O when fetching min and max for keyword fields
+area: Search
+type: bug
+issues: []

server/src/main/java/org/elasticsearch/search/internal/ExitableDirectoryReader.java

@@ -173,7 +173,7 @@ class ExitableDirectoryReader extends FilterDirectoryReader {
 
         private final QueryCancellation queryCancellation;
 
-        private ExitableTerms(Terms terms, QueryCancellation queryCancellation) {
+        ExitableTerms(Terms terms, QueryCancellation queryCancellation) {
             super(terms);
             this.queryCancellation = queryCancellation;
         }
@@ -187,6 +187,16 @@ class ExitableDirectoryReader extends FilterDirectoryReader {
         public TermsEnum iterator() throws IOException {
             return new ExitableTermsEnum(in.iterator(), queryCancellation);
         }
+
+        @Override
+        public BytesRef getMin() throws IOException {
+            return in.getMin();
+        }
+
+        @Override
+        public BytesRef getMax() throws IOException {
+            return in.getMax();
+        }
     }
 
     /**

server/src/main/java/org/elasticsearch/search/internal/FieldUsageTrackingDirectoryReader.java

@@ -90,11 +90,11 @@ public class FieldUsageTrackingDirectoryReader extends FilterDirectoryReader {
         void onKnnVectorsUsed(String field);
     }
 
-    public static final class FieldUsageTrackingLeafReader extends SequentialStoredFieldsLeafReader {
+    static final class FieldUsageTrackingLeafReader extends SequentialStoredFieldsLeafReader {
 
         private final FieldUsageNotifier notifier;
 
-        public FieldUsageTrackingLeafReader(LeafReader in, FieldUsageNotifier notifier) {
+        FieldUsageTrackingLeafReader(LeafReader in, FieldUsageNotifier notifier) {
             super(in);
             this.notifier = notifier;
         }
@@ -248,7 +248,7 @@ public class FieldUsageTrackingDirectoryReader extends FilterDirectoryReader {
             }
         }
 
-        private class FieldUsageTrackingTerms extends FilterTerms {
+        class FieldUsageTrackingTerms extends FilterTerms {
 
             private final String field;
 
@@ -286,6 +286,16 @@ public class FieldUsageTrackingDirectoryReader extends FilterDirectoryReader {
             public long getSumDocFreq() throws IOException {
                 return in.getSumDocFreq();
             }
+
+            @Override
+            public BytesRef getMin() throws IOException {
+                return in.getMin();
+            }
+
+            @Override
+            public BytesRef getMax() throws IOException {
+                return in.getMax();
+            }
         }
 
         private class FieldUsageTrackingTermsEnum extends FilterTermsEnum {

server/src/test/java/org/elasticsearch/search/internal/ContextIndexSearcherTests.java

@@ -16,6 +16,7 @@ import org.apache.lucene.document.IntPoint;
 import org.apache.lucene.document.StringField;
 import org.apache.lucene.index.DirectoryReader;
 import org.apache.lucene.index.FilterDirectoryReader;
+import org.apache.lucene.index.FilterLeafReader;
 import org.apache.lucene.index.IndexReader;
 import org.apache.lucene.index.IndexWriter;
 import org.apache.lucene.index.IndexWriterConfig;
@@ -24,6 +25,7 @@ import org.apache.lucene.index.LeafReaderContext;
 import org.apache.lucene.index.NoMergePolicy;
 import org.apache.lucene.index.PostingsEnum;
 import org.apache.lucene.index.Term;
+import org.apache.lucene.index.Terms;
 import org.apache.lucene.index.TermsEnum;
 import org.apache.lucene.search.BoostQuery;
 import org.apache.lucene.search.BulkScorer;
@@ -271,6 +273,43 @@ public class ContextIndexSearcherTests extends ESTestCase {
         IOUtils.close(reader, w, dir);
     }
 
+    public void testExitableTermsMinAndMax() throws IOException {
+        Directory dir = newDirectory();
+        IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(null));
+        Document doc = new Document();
+        StringField fooField = new StringField("foo", "bar", Field.Store.NO);
+        doc.add(fooField);
+        w.addDocument(doc);
+        w.flush();
+
+        DirectoryReader directoryReader = DirectoryReader.open(w);
+        for (LeafReaderContext lfc : directoryReader.leaves()) {
+            Terms terms = lfc.reader().terms("foo");
+            FilterLeafReader.FilterTerms filterTerms = new ExitableTerms(terms, new ExitableDirectoryReader.QueryCancellation() {
+                @Override
+                public boolean isEnabled() {
+                    return false;
+                }
+
+                @Override
+                public void checkCancelled() {
+
+                }
+            }) {
+                @Override
+                public TermsEnum iterator() {
+                    fail("Retrieving min and max should retrieve values from block tree instead of iterating");
+                    return null;
+                }
+            };
+            assertEquals("bar", filterTerms.getMin().utf8ToString());
+            assertEquals("bar", filterTerms.getMax().utf8ToString());
+        }
+        w.close();
+        directoryReader.close();
+        dir.close();
+    }
+
     private SparseFixedBitSet query(LeafReaderContext leaf, String field, String value) throws IOException {
         SparseFixedBitSet sparseFixedBitSet = new SparseFixedBitSet(leaf.reader().maxDoc());
         TermsEnum tenum = leaf.reader().terms(field).iterator();

server/src/test/java/org/elasticsearch/search/internal/FieldUsageTrackingDirectoryReaderTests.java

@@ -0,0 +1,115 @@
+/*
+ * 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 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 or the Server
+ * Side Public License, v 1.
+ */
+
+package org.elasticsearch.search.internal;
+
+import org.apache.lucene.document.Document;
+import org.apache.lucene.document.Field;
+import org.apache.lucene.document.StringField;
+import org.apache.lucene.index.DirectoryReader;
+import org.apache.lucene.index.IndexWriter;
+import org.apache.lucene.index.LeafReaderContext;
+import org.apache.lucene.index.TermsEnum;
+import org.apache.lucene.store.Directory;
+import org.elasticsearch.test.ESTestCase;
+
+import java.io.IOException;
+
+public class FieldUsageTrackingDirectoryReaderTests extends ESTestCase {
+
+    public void testTermsMinAndMax() throws IOException {
+        Directory dir = newDirectory();
+        IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(null));
+        Document doc = new Document();
+        StringField fooField = new StringField("foo", "bar", Field.Store.NO);
+        doc.add(fooField);
+        w.addDocument(doc);
+        w.flush();
+
+        DirectoryReader directoryReader = DirectoryReader.open(w);
+        for (LeafReaderContext lrc : directoryReader.leaves()) {
+            FieldUsageTrackingDirectoryReader.FieldUsageTrackingLeafReader leafReader =
+                new FieldUsageTrackingDirectoryReader.FieldUsageTrackingLeafReader(lrc.reader(), new TestFieldUsageNotifier());
+            FieldUsageTrackingDirectoryReader.FieldUsageTrackingLeafReader.FieldUsageTrackingTerms terms =
+                leafReader.new FieldUsageTrackingTerms("foo", lrc.reader().terms("foo")) {
+                    @Override
+                    public TermsEnum iterator() {
+                        fail("Retrieving min and max should retrieve values from block tree instead of iterating");
+                        return null;
+                    }
+                };
+            assertEquals("bar", terms.getMin().utf8ToString());
+            assertEquals("bar", terms.getMax().utf8ToString());
+        }
+        w.close();
+        directoryReader.close();
+        dir.close();
+    }
+
+    private static class TestFieldUsageNotifier implements FieldUsageTrackingDirectoryReader.FieldUsageNotifier {
+        @Override
+        public void onTermsUsed(String field) {
+
+        }
+
+        @Override
+        public void onPostingsUsed(String field) {
+
+        }
+
+        @Override
+        public void onTermFrequenciesUsed(String field) {
+
+        }
+
+        @Override
+        public void onPositionsUsed(String field) {
+
+        }
+
+        @Override
+        public void onOffsetsUsed(String field) {
+
+        }
+
+        @Override
+        public void onDocValuesUsed(String field) {
+
+        }
+
+        @Override
+        public void onStoredFieldsUsed(String field) {
+
+        }
+
+        @Override
+        public void onNormsUsed(String field) {
+
+        }
+
+        @Override
+        public void onPayloadsUsed(String field) {
+
+        }
+
+        @Override
+        public void onPointsUsed(String field) {
+
+        }
+
+        @Override
+        public void onTermVectorsUsed(String field) {
+
+        }
+
+        @Override
+        public void onKnnVectorsUsed(String field) {
+
+        }
+    }
+}