Use reader attributes to control term dict memory useage (#42838)

This change makes use of the reader attributes added in LUCENE-8671
to ensure that `_id` fields are always on-heap for best update performance
and term dicts are generally off-heap on Read-Only engines.

Closes #38390

Author: s1monw

server/src/main/java/org/elasticsearch/index/engine/InternalEngine.java

@@ -20,6 +20,8 @@
 package org.elasticsearch.index.engine;
 
 import org.apache.logging.log4j.Logger;
+import org.apache.lucene.codecs.blocktree.BlockTreeTermsReader;
+import org.apache.lucene.codecs.blocktree.BlockTreeTermsReader.FSTLoadMode;
 import org.apache.lucene.document.Field;
 import org.apache.lucene.document.NumericDocValuesField;
 import org.apache.lucene.index.DirectoryReader;
@@ -42,7 +44,9 @@
 import org.apache.lucene.search.TermQuery;
 import org.apache.lucene.store.AlreadyClosedException;
 import org.apache.lucene.store.Directory;
+import org.apache.lucene.store.FilterDirectory;
 import org.apache.lucene.store.LockObtainFailedException;
+import org.apache.lucene.store.MMapDirectory;
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.InfoStream;
 import org.elasticsearch.Assertions;
@@ -77,6 +81,7 @@
 import org.elasticsearch.index.seqno.SequenceNumbers;
 import org.elasticsearch.index.shard.ElasticsearchMergePolicy;
 import org.elasticsearch.index.shard.ShardId;
+import org.elasticsearch.index.store.FsDirectoryFactory;
 import org.elasticsearch.index.store.Store;
 import org.elasticsearch.index.translog.Translog;
 import org.elasticsearch.index.translog.TranslogConfig;
@@ -2140,10 +2145,21 @@ IndexWriter createWriter(Directory directory, IndexWriterConfig iwc) throws IOEx
         }
     }
 
+    static Map<String, String> getReaderAttributes(Directory directory) {
+        Directory unwrap = FilterDirectory.unwrap(directory);
+        boolean defaultOffHeap = FsDirectoryFactory.isHybridFs(unwrap) || unwrap instanceof MMapDirectory;
+        return Map.of(
+            BlockTreeTermsReader.FST_MODE_KEY, // if we are using MMAP for term dics we force all off heap unless it's the ID field
+            defaultOffHeap ? FSTLoadMode.OFF_HEAP.name() : FSTLoadMode.ON_HEAP.name()
+            , BlockTreeTermsReader.FST_MODE_KEY + "." + IdFieldMapper.NAME, // always force ID field on-heap for fast updates
+            FSTLoadMode.ON_HEAP.name());
+    }
+
     private IndexWriterConfig getIndexWriterConfig() {
         final IndexWriterConfig iwc = new IndexWriterConfig(engineConfig.getAnalyzer());
         iwc.setCommitOnClose(false); // we by default don't commit on close
         iwc.setOpenMode(IndexWriterConfig.OpenMode.APPEND);
+        iwc.setReaderAttributes(getReaderAttributes(store.directory()));
         iwc.setIndexDeletionPolicy(combinedDeletionPolicy);
         // with tests.verbose, lucene sets this up: plumb to align with filesystem stream
         boolean verbose = false;

server/src/main/java/org/elasticsearch/index/engine/NoOpEngine.java

@@ -46,7 +46,7 @@ public NoOpEngine(EngineConfig config) {
         super(config, null, null, true, Function.identity());
         this.stats = new SegmentsStats();
         Directory directory = store.directory();
-        try (DirectoryReader reader = DirectoryReader.open(directory)) {
+        try (DirectoryReader reader = DirectoryReader.open(directory, OFF_HEAP_READER_ATTRIBUTES)) {
             for (LeafReaderContext ctx : reader.getContext().leaves()) {
                 SegmentReader segmentReader = Lucene.segmentReader(ctx.reader());
                 fillSegmentStats(segmentReader, true, stats);

server/src/main/java/org/elasticsearch/index/engine/ReadOnlyEngine.java

@@ -18,6 +18,7 @@
  */
 package org.elasticsearch.index.engine;
 
+import org.apache.lucene.codecs.blocktree.BlockTreeTermsReader;
 import org.apache.lucene.index.DirectoryReader;
 import org.apache.lucene.index.IndexCommit;
 import org.apache.lucene.index.IndexReader;
@@ -47,7 +48,9 @@
 import java.io.IOException;
 import java.io.UncheckedIOException;
 import java.util.Arrays;
+import java.util.Collections;
 import java.util.List;
+import java.util.Map;
 import java.util.concurrent.CountDownLatch;
 import java.util.function.BiFunction;
 import java.util.function.Function;
@@ -62,6 +65,12 @@
  */
 public class ReadOnlyEngine extends Engine {
 
+    /**
+     * Reader attributes used for read only engines. These attributes prevent loading term dictionaries on-heap even if the field is an
+     * ID field.
+     */
+    public static final Map<String, String> OFF_HEAP_READER_ATTRIBUTES = Collections.singletonMap(BlockTreeTermsReader.FST_MODE_KEY,
+        BlockTreeTermsReader.FSTLoadMode.OFF_HEAP.name());
     private final SegmentInfos lastCommittedSegmentInfos;
     private final SeqNoStats seqNoStats;
     private final TranslogStats translogStats;
@@ -165,7 +174,7 @@ protected final DirectoryReader wrapReader(DirectoryReader reader,
     }
 
     protected DirectoryReader open(IndexCommit commit) throws IOException {
-        return DirectoryReader.open(commit);
+        return DirectoryReader.open(commit, OFF_HEAP_READER_ATTRIBUTES);
     }
 
     private DocsStats docsStats(final SegmentInfos lastCommittedSegmentInfos) {

server/src/main/java/org/elasticsearch/index/store/FsDirectoryFactory.java

@@ -22,6 +22,7 @@
 import org.apache.lucene.store.Directory;
 import org.apache.lucene.store.FSDirectory;
 import org.apache.lucene.store.FileSwitchDirectory;
+import org.apache.lucene.store.FilterDirectory;
 import org.apache.lucene.store.IOContext;
 import org.apache.lucene.store.IndexInput;
 import org.apache.lucene.store.LockFactory;
@@ -121,6 +122,14 @@ public String[] listAll() throws IOException {
         return directory;
     }
 
+    /**
+     * Returns true iff the directory is a hybrid fs directory
+     */
+    public static boolean isHybridFs(Directory directory) {
+        Directory unwrap = FilterDirectory.unwrap(directory);
+        return unwrap instanceof HybridDirectory;
+    }
+
     static final class HybridDirectory extends NIOFSDirectory {
         private final FSDirectory randomAccessDirectory;
 

server/src/main/java/org/elasticsearch/index/store/Store.java

@@ -38,15 +38,13 @@
 import org.apache.lucene.store.AlreadyClosedException;
 import org.apache.lucene.store.BufferedChecksum;
 import org.apache.lucene.store.ByteArrayDataInput;
-import org.apache.lucene.store.ByteBufferIndexInput;
 import org.apache.lucene.store.ChecksumIndexInput;
 import org.apache.lucene.store.Directory;
 import org.apache.lucene.store.FilterDirectory;
 import org.apache.lucene.store.IOContext;
 import org.apache.lucene.store.IndexInput;
 import org.apache.lucene.store.IndexOutput;
 import org.apache.lucene.store.Lock;
-import org.apache.lucene.store.RandomAccessInput;
 import org.apache.lucene.store.SimpleFSDirectory;
 import org.apache.lucene.util.ArrayUtil;
 import org.apache.lucene.util.BytesRef;
@@ -98,7 +96,6 @@
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
-import java.util.Set;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.concurrent.locks.ReentrantReadWriteLock;
@@ -137,7 +134,7 @@ public class Store extends AbstractIndexShardComponent implements Closeable, Ref
      * this by exploiting lucene internals and wrapping the IndexInput in a simple delegate.
      */
     public static final Setting<Boolean> FORCE_RAM_TERM_DICT = Setting.boolSetting("index.force_memory_term_dictionary", false,
-        Property.IndexScope);
+        Property.IndexScope, Property.Deprecated);
     static final String CODEC = "store";
     static final int VERSION_WRITE_THROWABLE= 2; // we write throwable since 2.0
     static final int VERSION_STACK_TRACE = 1; // we write the stack trace too since 1.4.0
@@ -172,8 +169,7 @@ public Store(ShardId shardId, IndexSettings indexSettings, Directory directory,
         final TimeValue refreshInterval = indexSettings.getValue(INDEX_STORE_STATS_REFRESH_INTERVAL_SETTING);
         logger.debug("store stats are refreshed with refresh_interval [{}]", refreshInterval);
         ByteSizeCachingDirectory sizeCachingDir = new ByteSizeCachingDirectory(directory, refreshInterval);
-        this.directory = new StoreDirectory(sizeCachingDir, Loggers.getLogger("index.store.deletes", shardId),
-            indexSettings.getValue(FORCE_RAM_TERM_DICT));
+        this.directory = new StoreDirectory(sizeCachingDir, Loggers.getLogger("index.store.deletes", shardId));
         this.shardLock = shardLock;
         this.onClose = onClose;
 
@@ -712,12 +708,10 @@ public int refCount() {
     static final class StoreDirectory extends FilterDirectory {
 
         private final Logger deletesLogger;
-        private final boolean forceRamTermDict;
 
-        StoreDirectory(ByteSizeCachingDirectory delegateDirectory, Logger deletesLogger, boolean forceRamTermDict) {
+        StoreDirectory(ByteSizeCachingDirectory delegateDirectory, Logger deletesLogger) {
             super(delegateDirectory);
             this.deletesLogger = deletesLogger;
-            this.forceRamTermDict = forceRamTermDict;
         }
 
         /** Estimate the cumulative size of all files in this directory in bytes. */
@@ -744,18 +738,6 @@ private void innerClose() throws IOException {
             super.close();
         }
 
-        @Override
-        public IndexInput openInput(String name, IOContext context) throws IOException {
-            IndexInput input = super.openInput(name, context);
-            if (name.endsWith(".tip") || name.endsWith(".cfs")) {
-                // only do this if we are reading cfs or tip file - all other files don't need this.
-                if (forceRamTermDict && input instanceof ByteBufferIndexInput) {
-                    return new DeoptimizingIndexInput(input.toString(), input);
-                }
-            }
-            return input;
-        }
-
         @Override
         public String toString() {
             return "store(" + in.toString() + ")";
@@ -1614,127 +1596,4 @@ private static IndexWriterConfig newIndexWriterConfig() {
                 // we also don't specify a codec here and merges should use the engines for this index
                 .setMergePolicy(NoMergePolicy.INSTANCE);
     }
-
-    /**
-     * see {@link #FORCE_RAM_TERM_DICT} for details
-     */
-    private static final class DeoptimizingIndexInput extends IndexInput {
-
-        private final IndexInput in;
-
-        private DeoptimizingIndexInput(String resourceDescription, IndexInput in) {
-            super(resourceDescription);
-            this.in = in;
-        }
-
-        @Override
-        public IndexInput clone() {
-            return new DeoptimizingIndexInput(toString(), in.clone());
-        }
-
-        @Override
-        public void close() throws IOException {
-            in.close();
-        }
-
-        @Override
-        public long getFilePointer() {
-            return in.getFilePointer();
-        }
-
-        @Override
-        public void seek(long pos) throws IOException {
-            in.seek(pos);
-        }
-
-        @Override
-        public long length() {
-            return in.length();
-        }
-
-        @Override
-        public String toString() {
-            return in.toString();
-        }
-
-        @Override
-        public IndexInput slice(String sliceDescription, long offset, long length) throws IOException {
-            return new DeoptimizingIndexInput(sliceDescription, in.slice(sliceDescription, offset, length));
-        }
-
-        @Override
-        public RandomAccessInput randomAccessSlice(long offset, long length) throws IOException {
-            return in.randomAccessSlice(offset, length);
-        }
-
-        @Override
-        public byte readByte() throws IOException {
-            return in.readByte();
-        }
-
-        @Override
-        public void readBytes(byte[] b, int offset, int len) throws IOException {
-            in.readBytes(b, offset, len);
-        }
-
-        @Override
-        public void readBytes(byte[] b, int offset, int len, boolean useBuffer) throws IOException {
-            in.readBytes(b, offset, len, useBuffer);
-        }
-
-        @Override
-        public short readShort() throws IOException {
-            return in.readShort();
-        }
-
-        @Override
-        public int readInt() throws IOException {
-            return in.readInt();
-        }
-
-        @Override
-        public int readVInt() throws IOException {
-            return in.readVInt();
-        }
-
-        @Override
-        public int readZInt() throws IOException {
-            return in.readZInt();
-        }
-
-        @Override
-        public long readLong() throws IOException {
-            return in.readLong();
-        }
-
-        @Override
-        public long readVLong() throws IOException {
-            return in.readVLong();
-        }
-
-        @Override
-        public long readZLong() throws IOException {
-            return in.readZLong();
-        }
-
-        @Override
-        public String readString() throws IOException {
-            return in.readString();
-        }
-
-        @Override
-        public Map<String, String> readMapOfStrings() throws IOException {
-            return in.readMapOfStrings();
-        }
-
-        @Override
-        public Set<String> readSetOfStrings() throws IOException {
-            return in.readSetOfStrings();
-        }
-
-        @Override
-        public void skipBytes(long numBytes) throws IOException {
-            in.skipBytes(numBytes);
-        }
-    }
 }