triedb/pathdb: introduce metadata and version tracking

core/rawdb/accessors_history.go

@@ -17,35 +17,29 @@
 package rawdb
 
 import (
-	"encoding/binary"
-
 	"github.com/ethereum/go-ethereum/common"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/log"
 )
 
-// ReadLastStateHistoryIndex retrieves the number of latest indexed state history.
+// ReadStateHistoryIndexMetadata retrieves the metadata of state history index.
-func ReadLastStateHistoryIndex(db ethdb.KeyValueReader) *uint64 {
+func ReadStateHistoryIndexMetadata(db ethdb.KeyValueReader) []byte {
 	data, _ := db.Get(headStateHistoryIndexKey)
-	if len(data) != 8 {
+	return data
-		return nil
-	}
-	number := binary.BigEndian.Uint64(data)
-	return &number
 }
 
-// WriteLastStateHistoryIndex stores the number of latest indexed state history
+// WriteStateHistoryIndexMetadata stores the metadata of state history index
 // into database.
-func WriteLastStateHistoryIndex(db ethdb.KeyValueWriter, number uint64) {
+func WriteStateHistoryIndexMetadata(db ethdb.KeyValueWriter, blob []byte) {
-	if err := db.Put(headStateHistoryIndexKey, encodeBlockNumber(number)); err != nil {
+	if err := db.Put(headStateHistoryIndexKey, blob); err != nil {
-		log.Crit("Failed to store the state index tail", "err", err)
+		log.Crit("Failed to store the metadata of state history index", "err", err)
 	}
 }
 
-// DeleteLastStateHistoryIndex removes the number of latest indexed state history.
+// DeleteStateHistoryIndexMetadata removes the metadata of state history index.
-func DeleteLastStateHistoryIndex(db ethdb.KeyValueWriter) {
+func DeleteStateHistoryIndexMetadata(db ethdb.KeyValueWriter) {
 	if err := db.Delete(headStateHistoryIndexKey); err != nil {
-		log.Crit("Failed to delete the state index tail", "err", err)
+		log.Crit("Failed to delete the metadata of state history index", "err", err)
 	}
 }
 
@@ -158,12 +152,12 @@ func increaseKey(key []byte) []byte {
 	return nil
 }
 
-// DeleteHistoryIndex completely removes all history indexing data, including indexes
+// DeleteStateHistoryIndex completely removes all history indexing data, including
-// for accounts and storages.
+// indexes for accounts and storages.
 //
 // Note, this method assumes the storage space with prefix `StateHistoryIndexPrefix`
 // is exclusively occupied by the history indexing data!
-func DeleteHistoryIndex(db ethdb.KeyValueRangeDeleter) {
+func DeleteStateHistoryIndex(db ethdb.KeyValueRangeDeleter) {
 	if err := db.DeleteRange(StateHistoryIndexPrefix, increaseKey(StateHistoryIndexPrefix)); err != nil {
 		log.Crit("Failed to delete history index range", "err", err)
 	}

triedb/pathdb/database.go

@@ -314,8 +314,8 @@ func (db *Database) repairHistory() error {
 			// TODO(rjl493456442) would be better to group them into a batch.
 			//
 			// Purge all state history indexing data first
-			rawdb.DeleteLastStateHistoryIndex(db.diskdb)
+			rawdb.DeleteStateHistoryIndexMetadata(db.diskdb)
-			rawdb.DeleteHistoryIndex(db.diskdb)
+			rawdb.DeleteStateHistoryIndex(db.diskdb)
 			err := db.freezer.Reset()
 			if err != nil {
 				log.Crit("Failed to reset state histories", "err", err)
@@ -507,8 +507,8 @@ func (db *Database) Enable(root common.Hash) error {
 		// TODO(rjl493456442) would be better to group them into a batch.
 		//
 		// Purge all state history indexing data first
-		rawdb.DeleteLastStateHistoryIndex(db.diskdb)
+		rawdb.DeleteStateHistoryIndexMetadata(db.diskdb)
-		rawdb.DeleteHistoryIndex(db.diskdb)
+		rawdb.DeleteStateHistoryIndex(db.diskdb)
 		if err := db.freezer.Reset(); err != nil {
 			return err
 		}

triedb/pathdb/history_index_test.go

@@ -189,8 +189,8 @@ func TestBatchIndexerWrite(t *testing.T) {
 	if err := batch.finish(true); err != nil {
 		t.Fatalf("Failed to finish batch indexer, %v", err)
 	}
-	indexed := rawdb.ReadLastStateHistoryIndex(db)
+	metadata := loadIndexMetadata(db)
-	if indexed == nil || *indexed != uint64(10) {
+	if metadata == nil || metadata.Last != uint64(10) {
 		t.Fatal("Unexpected index position")
 	}
 	var (
@@ -278,8 +278,8 @@ func TestBatchIndexerDelete(t *testing.T) {
 		t.Fatalf("Failed to finish batch indexer, %v", err)
 	}
 
-	indexed := rawdb.ReadLastStateHistoryIndex(db)
+	metadata := loadIndexMetadata(db)
-	if indexed != nil {
+	if metadata != nil {
 		t.Fatal("Unexpected index position")
 	}
 	it := db.NewIterator(rawdb.StateHistoryIndexPrefix, nil)

triedb/pathdb/history_indexer.go

@@ -28,10 +28,44 @@ import (
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/log"
+	"github.com/ethereum/go-ethereum/rlp"
 )
 
-// The batch size for reading state histories
+const (
-const historyReadBatch = 1000
+	// The batch size for reading state histories
+	historyReadBatch = 1000
+
+	stateIndexV0      = uint8(0)     // initial version of state index structure
+	stateIndexVersion = stateIndexV0 // the current state index version
+)
+
+type indexMetadata struct {
+	Version uint8
+	Last    uint64
+}
+
+func loadIndexMetadata(db ethdb.KeyValueReader) *indexMetadata {
+	blob := rawdb.ReadStateHistoryIndexMetadata(db)
+	if len(blob) == 0 {
+		return nil
+	}
+	var m indexMetadata
+	if err := rlp.DecodeBytes(blob, &m); err != nil {
+		return nil
+	}
+	return &m
+}
+
+func storeIndexMetadata(db ethdb.KeyValueWriter, last uint64) {
+	var m indexMetadata
+	m.Version = stateIndexVersion
+	m.Last = last
+	blob, err := rlp.EncodeToBytes(m)
+	if err != nil {
+		log.Crit("Failed to encode index metadata", "err", err)
+	}
+	rawdb.WriteStateHistoryIndexMetadata(db, blob)
+}
 
 // batchIndexer is a structure designed to perform batch indexing or unindexing
 // of state histories atomically.
@@ -144,12 +178,12 @@ func (b *batchIndexer) finish(force bool) error {
 	}
 	// Update the position of last indexed state history
 	if !b.delete {
-		rawdb.WriteLastStateHistoryIndex(batch, b.lastID)
+		storeIndexMetadata(batch, b.lastID)
 	} else {
 		if b.lastID == 1 {
-			rawdb.DeleteLastStateHistoryIndex(batch)
+			rawdb.DeleteStateHistoryIndexMetadata(batch)
 		} else {
-			rawdb.WriteLastStateHistoryIndex(batch, b.lastID-1)
+			storeIndexMetadata(batch, b.lastID-1)
 		}
 	}
 	if err := batch.Write(); err != nil {
@@ -167,11 +201,11 @@ func indexSingle(historyID uint64, db ethdb.KeyValueStore, freezer ethdb.Ancient
 		indexHistoryTimer.UpdateSince(start)
 	}(time.Now())
 
-	indexed := rawdb.ReadLastStateHistoryIndex(db)
+	metadata := loadIndexMetadata(db)
-	if indexed == nil || *indexed+1 != historyID {
+	if metadata == nil || metadata.Last+1 != historyID {
 		last := "null"
-		if indexed != nil {
+		if metadata != nil {
-			last = fmt.Sprintf("%v", *indexed)
+			last = fmt.Sprintf("%v", metadata.Last)
 		}
 		return fmt.Errorf("history indexing is out of order, last: %s, requested: %d", last, historyID)
 	}
@@ -196,11 +230,11 @@ func unindexSingle(historyID uint64, db ethdb.KeyValueStore, freezer ethdb.Ancie
 		unindexHistoryTimer.UpdateSince(start)
 	}(time.Now())
 
-	indexed := rawdb.ReadLastStateHistoryIndex(db)
+	metadata := loadIndexMetadata(db)
-	if indexed == nil || *indexed != historyID {
+	if metadata == nil || metadata.Last != historyID {
 		last := "null"
-		if indexed != nil {
+		if metadata != nil {
-			last = fmt.Sprintf("%v", *indexed)
+			last = fmt.Sprintf("%v", metadata.Last)
 		}
 		return fmt.Errorf("history unindexing is out of order, last: %s, requested: %d", last, historyID)
 	}
@@ -286,8 +320,8 @@ func (i *indexIniter) run(lastID uint64) {
 		// checkDone indicates whether all requested state histories
 		// have been fully indexed.
 		checkDone = func() bool {
-			indexed := rawdb.ReadLastStateHistoryIndex(i.disk)
+			metadata := loadIndexMetadata(i.disk)
-			return indexed != nil && *indexed == lastID
+			return metadata != nil && metadata.Last == lastID
 		}
 	)
 	go i.index(done, interrupt, lastID)
@@ -364,22 +398,22 @@ func (i *indexIniter) next() (uint64, error) {
 	tailID := tail + 1 // compute the id of the oldest history
 
 	// Start indexing from scratch if nothing has been indexed
-	lastIndexed := rawdb.ReadLastStateHistoryIndex(i.disk)
+	metadata := loadIndexMetadata(i.disk)
-	if lastIndexed == nil {
+	if metadata == nil {
 		log.Debug("Initialize state history indexing from scratch", "id", tailID)
 		return tailID, nil
 	}
 	// Resume indexing from the last interrupted position
-	if *lastIndexed+1 >= tailID {
+	if metadata.Last+1 >= tailID {
-		log.Debug("Resume state history indexing", "id", *lastIndexed+1, "tail", tailID)
+		log.Debug("Resume state history indexing", "id", metadata.Last+1, "tail", tailID)
-		return *lastIndexed + 1, nil
+		return metadata.Last + 1, nil
 	}
 	// History has been shortened without indexing. Discard the gapped segment
 	// in the history and shift to the first available element.
 	//
 	// The missing indexes corresponding to the gapped histories won't be visible.
 	// It's fine to leave them unindexed.
-	log.Info("History gap detected, discard old segment", "oldHead", *lastIndexed, "newHead", tailID)
+	log.Info("History gap detected, discard old segment", "oldHead", metadata.Last, "newHead", tailID)
 	return tailID, nil
 }
 
@@ -476,6 +510,14 @@ type historyIndexer struct {
 // newHistoryIndexer constructs the history indexer and launches the background
 // initer to complete the indexing of any remaining state histories.
 func newHistoryIndexer(disk ethdb.KeyValueStore, freezer ethdb.AncientStore, lastHistoryID uint64) *historyIndexer {
+	// Purge the obsolete index data from the database.
+	metadata := loadIndexMetadata(disk)
+	if metadata != nil && metadata.Version != stateIndexVersion {
+		// TODO(rjl493456442) would be better to group them into a batch.
+		rawdb.DeleteStateHistoryIndexMetadata(disk)
+		rawdb.DeleteStateHistoryIndex(disk)
+		log.Info("Cleaned up obsolete state history index", "version", metadata.Version, "want", stateIndexVersion)
+	}
 	return &historyIndexer{
 		initer:  newIndexIniter(disk, freezer, lastHistoryID),
 		disk:    disk,

triedb/pathdb/history_reader.go

@@ -113,8 +113,8 @@ type indexReaderWithLimitTag struct {
 // newIndexReaderWithLimitTag constructs a index reader with indexing position.
 func newIndexReaderWithLimitTag(db ethdb.KeyValueReader, state stateIdent) (*indexReaderWithLimitTag, error) {
 	// Read the last indexed ID before the index reader construction
-	indexed := rawdb.ReadLastStateHistoryIndex(db)
+	metadata := loadIndexMetadata(db)
-	if indexed == nil {
+	if metadata == nil {
 		return nil, errors.New("state history hasn't been indexed yet")
 	}
 	r, err := newIndexReader(db, state)
@@ -123,7 +123,7 @@ func newIndexReaderWithLimitTag(db ethdb.KeyValueReader, state stateIdent) (*ind
 	}
 	return &indexReaderWithLimitTag{
 		reader: r,
-		limit:  *indexed,
+		limit:  metadata.Last,
 		db:     db,
 	}, nil
 }
@@ -157,14 +157,14 @@ func (r *indexReaderWithLimitTag) readGreaterThan(id uint64, lastID uint64) (uin
 		return res, nil
 	}
 	// Refresh the index reader and give another attempt
-	indexed := rawdb.ReadLastStateHistoryIndex(r.db)
+	metadata := loadIndexMetadata(r.db)
-	if indexed == nil || *indexed < lastID {
+	if metadata == nil || metadata.Last < lastID {
 		return 0, errors.New("state history hasn't been indexed yet")
 	}
 	if err := r.reader.refresh(); err != nil {
 		return 0, err
 	}
-	r.limit = *indexed
+	r.limit = metadata.Last
 
 	return r.reader.readGreaterThan(id)
 }
@@ -308,15 +308,15 @@ func (r *historyReader) read(state stateIdentQuery, stateID uint64, lastID uint6
 	if stateID < tail {
 		return nil, errors.New("historical state has been pruned")
 	}
-	lastIndexedID := rawdb.ReadLastStateHistoryIndex(r.disk)
 
 	// To serve the request, all state histories from stateID+1 to lastID
 	// must be indexed. It's not supposed to happen unless system is very
 	// wrong.
-	if lastIndexedID == nil || *lastIndexedID < lastID {
+	metadata := loadIndexMetadata(r.disk)
+	if metadata == nil || metadata.Last < lastID {
 		indexed := "null"
-		if lastIndexedID != nil {
+		if metadata != nil {
-			indexed = fmt.Sprintf("%d", *lastIndexedID)
+			indexed = fmt.Sprintf("%d", metadata.Last)
 		}
 		return nil, fmt.Errorf("state history is not fully indexed, requested: %d, indexed: %s", stateID, indexed)
 	}

triedb/pathdb/history_reader_test.go

@@ -28,8 +28,8 @@ import (
 
 func waitIndexing(db *Database) {
 	for {
-		id := rawdb.ReadLastStateHistoryIndex(db.diskdb)
+		metadata := loadIndexMetadata(db.diskdb)
-		if id != nil && *id >= db.tree.bottom().stateID() {
+		if metadata != nil && metadata.Last >= db.tree.bottom().stateID() {
 			return
 		}
 		time.Sleep(100 * time.Millisecond)