feat(blockdb): add lru cache for block entries

Author: DracoLi

x/blockdb/README.md

@@ -10,6 +10,7 @@ BlockDB is a specialized database optimized for blockchain blocks.
 - **Configurable Durability**: Optional `syncToDisk` mode guarantees immediate recoverability
 - **Automatic Recovery**: Detects and recovers unindexed blocks after unclean shutdowns
 - **Block Compression**: zstd compression for block data
+- **In-Memory Cache**: LRU cache for recently accessed blocks
 
 ## Design
 
@@ -167,7 +168,6 @@ if err != nil {
 
 ## TODO
 
-- Implement a block cache for recently accessed blocks
 - Use a buffered pool to avoid allocations on reads and writes
 - Add performance benchmarks
 - Consider supporting missing data files (currently we error if any data files are missing)

x/blockdb/config.go

@@ -11,6 +11,9 @@ const DefaultMaxDataFileSize = 500 * 1024 * 1024 * 1024
 // DefaultMaxDataFiles is the default maximum number of data files descriptors cached.
 const DefaultMaxDataFiles = 10
 
+// DefaultEntryCacheSize is the default size of the entry cache.
+const DefaultEntryCacheSize = 256
+
 // DatabaseConfig contains configuration parameters for BlockDB.
 type DatabaseConfig struct {
 	// IndexDir is the directory where the index file is stored.
@@ -28,6 +31,9 @@ type DatabaseConfig struct {
 	// MaxDataFiles is the maximum number of data files descriptors cached.
 	MaxDataFiles int
 
+	// EntryCacheSize is the size of the entry cache (default: 256).
+	EntryCacheSize int
+
 	// CheckpointInterval defines how frequently (in blocks) the index file header is updated (default: 1024).
 	CheckpointInterval uint64
 
@@ -43,6 +49,7 @@ func DefaultConfig() DatabaseConfig {
 		MinimumHeight:      0,
 		MaxDataFileSize:    DefaultMaxDataFileSize,
 		MaxDataFiles:       DefaultMaxDataFiles,
+		EntryCacheSize:     DefaultEntryCacheSize,
 		CheckpointInterval: 1024,
 		SyncToDisk:         true,
 	}
@@ -91,6 +98,12 @@ func (c DatabaseConfig) WithMaxDataFiles(maxFiles int) DatabaseConfig {
 	return c
 }
 
+// WithEntryCacheSize returns a copy of the config with EntryCacheSize set to the given value.
+func (c DatabaseConfig) WithEntryCacheSize(size int) DatabaseConfig {
+	c.EntryCacheSize = size
+	return c
+}
+
 // WithCheckpointInterval returns a copy of the config with CheckpointInterval set to the given value.
 func (c DatabaseConfig) WithCheckpointInterval(interval uint64) DatabaseConfig {
 	c.CheckpointInterval = interval
@@ -114,5 +127,8 @@ func (c DatabaseConfig) Validate() error {
 	if c.MaxDataFileSize == 0 {
 		return errors.New("MaxDataFileSize must be positive")
 	}
+	if c.EntryCacheSize < 1 {
+		return errors.New("EntryCacheSize cannot be less than 1")
+	}
 	return nil
 }

x/blockdb/database.go

@@ -12,6 +12,7 @@ import (
 	"math"
 	"os"
 	"path/filepath"
+	"slices"
 	"sync"
 	"sync/atomic"
 
@@ -176,6 +177,7 @@ type Database struct {
 	log        logging.Logger
 	closed     bool
 	fileCache  *lru.Cache[int, *os.File]
+	entryCache *lru.Cache[BlockHeight, BlockData]
 	compressor compression.Compressor
 
 	// closeMu prevents the database from being closed while in use and prevents
@@ -223,6 +225,7 @@ func New(config DatabaseConfig, log logging.Logger) (*Database, error) {
 				f.Close()
 			}
 		}),
+		entryCache: lru.NewCache[BlockHeight, BlockData](config.EntryCacheSize),
 		compressor: compressor,
 	}
 
@@ -231,6 +234,7 @@ func New(config DatabaseConfig, log logging.Logger) (*Database, error) {
 		zap.String("dataDir", config.DataDir),
 		zap.Uint64("maxDataFileSize", config.MaxDataFileSize),
 		zap.Int("maxDataFiles", config.MaxDataFiles),
+		zap.Int("entryCacheSize", config.EntryCacheSize),
 	)
 
 	if err := s.openAndInitializeIndex(); err != nil {
@@ -275,6 +279,7 @@ func (s *Database) Close() error {
 	}
 
 	s.closeFiles()
+	s.entryCache.Flush()
 
 	s.log.Info("Block database closed successfully")
 	return err
@@ -371,6 +376,7 @@ func (s *Database) Put(height BlockHeight, block BlockData) error {
 		)
 		return err
 	}
+	s.entryCache.Put(height, slices.Clone(block))
 
 	s.log.Debug("Block written successfully",
 		zap.Uint64("height", height),
@@ -385,12 +391,6 @@ func (s *Database) Put(height BlockHeight, block BlockData) error {
 // It returns database.ErrNotFound if the block does not exist.
 func (s *Database) readBlockIndex(height BlockHeight) (indexEntry, error) {
 	var entry indexEntry
-	if s.closed {
-		s.log.Error("Failed to read block index: database is closed",
-			zap.Uint64("height", height),
-		)
-		return entry, database.ErrClosed
-	}
 
 	// Skip the index entry read if we know the block is past the max height.
 	maxHeight := s.maxBlockHeight.Load()
@@ -436,6 +436,15 @@ func (s *Database) Get(height BlockHeight) (BlockData, error) {
 	s.closeMu.RLock()
 	defer s.closeMu.RUnlock()
 
+	if s.closed {
+		s.log.Error("Failed Get: database closed", zap.Uint64("height", height))
+		return nil, database.ErrClosed
+	}
+
+	if c, ok := s.entryCache.Get(height); ok {
+		return slices.Clone(c), nil
+	}
+
 	indexEntry, err := s.readBlockIndex(height)
 	if err != nil {
 		return nil, err
@@ -486,6 +495,7 @@ func (s *Database) Get(height BlockHeight) (BlockData, error) {
 		return nil, fmt.Errorf("checksum mismatch: calculated %d, stored %d", calculatedChecksum, bh.Checksum)
 	}
 
+	s.entryCache.Put(height, slices.Clone(decompressed))
 	return decompressed, nil
 }
 
@@ -494,6 +504,14 @@ func (s *Database) Has(height BlockHeight) (bool, error) {
 	s.closeMu.RLock()
 	defer s.closeMu.RUnlock()
 
+	if s.closed {
+		s.log.Error("Failed Has: database closed", zap.Uint64("height", height))
+		return false, database.ErrClosed
+	}
+
+	if _, ok := s.entryCache.Get(height); ok {
+		return true, nil
+	}
 	_, err := s.readBlockIndex(height)
 	if err != nil {
 		if errors.Is(err, database.ErrNotFound) || errors.Is(err, ErrInvalidBlockHeight) {

x/blockdb/entry_cache_test.go

@@ -0,0 +1,107 @@
+// Copyright (C) 2019-2025, Ava Labs, Inc. All rights reserved.
+// See the file LICENSE for licensing terms.
+
+package blockdb
+
+import (
+	"slices"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestCacheOnMiss(t *testing.T) {
+	db, _ := newTestDatabase(t, DefaultConfig())
+	height := uint64(20)
+	block := randomBlock(t)
+	require.NoError(t, db.Put(height, block))
+
+	// Evict the entry from cache to simulate a cache miss
+	db.entryCache.Evict(height)
+
+	// Read the block - should populate the cache on cache miss
+	_, err := db.Get(height)
+	require.NoError(t, err)
+
+	_, ok := db.entryCache.Get(height)
+	require.True(t, ok)
+}
+
+func TestCacheHas(t *testing.T) {
+	db, _ := newTestDatabase(t, DefaultConfig())
+	height := uint64(30)
+	block := randomBlock(t)
+	require.NoError(t, db.Put(height, block))
+
+	has, err := db.Has(height)
+	require.NoError(t, err)
+	require.True(t, has)
+
+	// Verify block is in cache
+	cached, ok := db.entryCache.Get(height)
+	require.True(t, ok)
+	require.Equal(t, block, cached)
+}
+
+func TestCachePutStoresClone(t *testing.T) {
+	db, _ := newTestDatabase(t, DefaultConfig())
+	height := uint64(40)
+	block := randomBlock(t)
+	clone := slices.Clone(block)
+	require.NoError(t, db.Put(height, clone))
+
+	// Modify the original block after Put
+	clone[0] = 99
+
+	// Cache should have the original unmodified data
+	cached, ok := db.entryCache.Get(height)
+	require.True(t, ok)
+	require.Equal(t, block, cached)
+}
+
+func TestCacheGetReturnsClone(t *testing.T) {
+	db, _ := newTestDatabase(t, DefaultConfig())
+	height := uint64(50)
+	block := randomBlock(t)
+	require.NoError(t, db.Put(height, block))
+
+	// Get the block and modify the returned data
+	data, err := db.Get(height)
+	require.NoError(t, err)
+	data[0] = 99
+
+	// Cache should still have the original unmodified data
+	cached, ok := db.entryCache.Get(height)
+	require.True(t, ok)
+	require.Equal(t, block, cached)
+
+	// Second Get should also return original data
+	data, err = db.Get(height)
+	require.NoError(t, err)
+	require.Equal(t, block, data)
+}
+
+func TestCachePutOverridesSameHeight(t *testing.T) {
+	db, _ := newTestDatabase(t, DefaultConfig())
+	height := uint64(60)
+	b1 := randomBlock(t)
+	require.NoError(t, db.Put(height, b1))
+
+	// Verify first block is in cache
+	cached, ok := db.entryCache.Get(height)
+	require.True(t, ok)
+	require.Equal(t, b1, cached)
+
+	// Put second block at same height and verify it overrides the first one
+	b2 := randomBlock(t)
+	require.NoError(t, db.Put(height, b2))
+	cached, ok = db.entryCache.Get(height)
+	require.True(t, ok)
+	require.Equal(t, b2, cached)
+	require.NotEqual(t, b1, cached)
+
+	// Get should also return the new block
+	data, err := db.Get(height)
+	require.NoError(t, err)
+	require.Equal(t, b2, data)
+}

x/blockdb/readblock_test.go

@@ -51,6 +51,7 @@ func TestReadOperations(t *testing.T) {
 				MaxDataFileSize:    DefaultMaxDataFileSize,
 				CheckpointInterval: 1024,
 				MaxDataFiles:       DefaultMaxDataFileSize,
+				EntryCacheSize:     DefaultEntryCacheSize,
 			},
 		},
 		{
@@ -69,6 +70,7 @@ func TestReadOperations(t *testing.T) {
 				MaxDataFileSize:    DefaultMaxDataFileSize,
 				CheckpointInterval: 1024,
 				MaxDataFiles:       DefaultMaxDataFileSize,
+				EntryCacheSize:     DefaultEntryCacheSize,
 			},
 			wantErr: ErrInvalidBlockHeight,
 		},

x/blockdb/writeblock_test.go

@@ -4,6 +4,7 @@
 package blockdb
 
 import (
+	"bytes"
 	"math"
 	"os"
 	"strings"
@@ -38,7 +39,7 @@ func TestPutGet(t *testing.T) {
 		{
 			name:  "nil block",
 			block: nil,
-			want:  []byte{},
+			want:  nil,
 		},
 	}
 	for _, tt := range tests {
@@ -49,7 +50,7 @@ func TestPutGet(t *testing.T) {
 
 			got, err := db.Get(0)
 			require.NoError(t, err)
-			require.Equal(t, tt.want, got)
+			require.True(t, bytes.Equal(tt.want, got))
 		})
 	}
 }