feat: lazy pruning

batch.go

@@ -51,13 +51,9 @@ func (b *BatchWithFlusher) Set(key, value []byte) error {
 		return err
 	}
 	if batchSizeAfter > b.flushThreshold {
-		if err := b.batch.Write(); err != nil {
+		if err := b.Write(); err != nil {
 			return err
 		}
-		if err := b.batch.Close(); err != nil {
-			return err
-		}
-		b.batch = b.db.NewBatchWithSize(b.flushThreshold)
 	}
 	return b.batch.Set(key, value)
 }
@@ -72,23 +68,33 @@ func (b *BatchWithFlusher) Delete(key []byte) error {
 		return err
 	}
 	if batchSizeAfter > b.flushThreshold {
-		if err := b.batch.Write(); err != nil {
-			return err
-		}
-		if err := b.batch.Close(); err != nil {
+		if err := b.Write(); err != nil {
 			return err
 		}
-		b.batch = b.db.NewBatchWithSize(b.flushThreshold)
 	}
 	return b.batch.Delete(key)
 }
 
 func (b *BatchWithFlusher) Write() error {
-	return b.batch.Write()
+	if err := b.batch.Write(); err != nil {
+		return err
+	}
+	if err := b.batch.Close(); err != nil {
+		return err
+	}
+	b.batch = b.db.NewBatchWithSize(b.flushThreshold)
+	return nil
 }
 
 func (b *BatchWithFlusher) WriteSync() error {
-	return b.batch.WriteSync()
+	if err := b.batch.WriteSync(); err != nil {
+		return err
+	}
+	if err := b.batch.Close(); err != nil {
+		return err
+	}
+	b.batch = b.db.NewBatchWithSize(b.flushThreshold)
+	return nil
 }
 
 func (b *BatchWithFlusher) Close() error {

benchmarks/bench_test.go

@@ -51,7 +51,7 @@ func commitTree(b *testing.B, t *iavl.MutableTree) {
 	}
 
 	if version > historySize {
-		err = t.DeleteVersionsTo(version - historySize)
+		err = t.DeleteVersionsToSync(version - historySize)
 		if err != nil {
 			b.Errorf("Can't delete: %v", err)
 		}

export_test.go

@@ -346,14 +346,14 @@ func TestExporter_DeleteVersionErrors(t *testing.T) {
 	require.NoError(t, err)
 	defer exporter.Close()
 
-	err = tree.DeleteVersionsTo(1)
+	err = tree.DeleteVersionsToSync(1)
 	require.NoError(t, err)
 
-	err = tree.DeleteVersionsTo(2)
+	err = tree.DeleteVersionsToSync(2)
 	require.Error(t, err)
 
 	exporter.Close()
-	err = tree.DeleteVersionsTo(2)
+	err = tree.DeleteVersionsToSync(2)
 	require.NoError(t, err)
 }
 

import.go

@@ -204,7 +204,7 @@ func (i *Importer) Commit() error {
 			return err
 		}
 		if i.stack[0].nodeKey.version < i.version { // it means there is no update in the given version
-			if err := i.batch.Set(i.tree.ndb.nodeKey(GetRootKey(i.version)), i.tree.ndb.nodeKeyPrefix(i.stack[0].nodeKey.version)); err != nil {
+			if err := i.batch.Set(i.tree.ndb.nodeKey(GetRootKey(i.version)), i.tree.ndb.nodeKey(i.stack[0].nodeKey.GetKey())); err != nil {
 				return err
 			}
 		}

iterator_test.go

@@ -179,7 +179,7 @@ func TestIterator_WithDelete_Full_Ascending_Success(t *testing.T) {
 	_, _, err = tree.SaveVersion()
 	require.NoError(t, err)
 
-	err = tree.DeleteVersionsTo(1)
+	err = tree.DeleteVersionsToSync(1)
 	require.NoError(t, err)
 
 	latestVersion, err := tree.ndb.getLatestVersion()

migrate_test.go

@@ -8,18 +8,20 @@ import (
 	"os/exec"
 	"path"
 	"testing"
+	"time"
 
 	"cosmossdk.io/log"
 	"github.com/stretchr/testify/require"
 
 	dbm "github.com/cosmos/iavl/db"
+	iavlrand "github.com/cosmos/iavl/internal/rand"
 )
 
 const (
 	dbType = "goleveldb"
 )
 
-func createLegacyTree(t *testing.T, dbType, dbDir string, version int) (string, error) {
+func createLegacyTree(t *testing.T, dbDir string, version int) (string, error) {
 	relateDir := path.Join(t.TempDir(), dbDir)
 	if _, err := os.Stat(relateDir); err == nil {
 		err := os.RemoveAll(relateDir)
@@ -48,10 +50,10 @@ func createLegacyTree(t *testing.T, dbType, dbDir string, version int) (string,
 func TestLazySet(t *testing.T) {
 	legacyVersion := 1000
 	dbDir := fmt.Sprintf("legacy-%s-%d", dbType, legacyVersion)
-	relateDir, err := createLegacyTree(t, dbType, dbDir, legacyVersion)
+	relateDir, err := createLegacyTree(t, dbDir, legacyVersion)
 	require.NoError(t, err)
 
-	db, err := dbm.NewDB("test", "goleveldb", relateDir)
+	db, err := dbm.NewDB("test", dbType, relateDir)
 	require.NoError(t, err)
 
 	defer func() {
@@ -91,10 +93,10 @@ func TestLazySet(t *testing.T) {
 func TestLegacyReferenceNode(t *testing.T) {
 	legacyVersion := 20
 	dbDir := fmt.Sprintf("./legacy-%s-%d", dbType, legacyVersion)
-	relateDir, err := createLegacyTree(t, dbType, dbDir, legacyVersion)
+	relateDir, err := createLegacyTree(t, dbDir, legacyVersion)
 	require.NoError(t, err)
 
-	db, err := dbm.NewDB("test", "goleveldb", relateDir)
+	db, err := dbm.NewDB("test", dbType, relateDir)
 	require.NoError(t, err)
 
 	defer func() {
@@ -123,17 +125,17 @@ func TestLegacyReferenceNode(t *testing.T) {
 	_, err = newTree.LoadVersion(version - 1)
 	require.NoError(t, err)
 	// Check if the reference node is refactored
-	require.Equal(t, newTree.root.nodeKey.nonce, uint32(1))
+	require.Equal(t, newTree.root.nodeKey.nonce, uint32(0))
 	require.Equal(t, newTree.root.nodeKey.version, int64(legacyVersion))
 }
 
 func TestDeleteVersions(t *testing.T) {
 	legacyVersion := 100
 	dbDir := fmt.Sprintf("./legacy-%s-%d", dbType, legacyVersion)
-	relateDir, err := createLegacyTree(t, dbType, dbDir, legacyVersion)
+	relateDir, err := createLegacyTree(t, dbDir, legacyVersion)
 	require.NoError(t, err)
 
-	db, err := dbm.NewDB("test", "goleveldb", relateDir)
+	db, err := dbm.NewDB("test", dbType, relateDir)
 	require.NoError(t, err)
 
 	defer func() {
@@ -195,13 +197,149 @@ func TestDeleteVersions(t *testing.T) {
 	}
 	// Check if the legacy versions are deleted at once
 	versions = tree.AvailableVersions()
-	err = tree.DeleteVersionsTo(legacyLatestVersion - 1)
+	err = tree.DeleteVersionsToSync(legacyLatestVersion - 1)
 	require.NoError(t, err)
 	pVersions := tree.AvailableVersions()
 	require.Equal(t, len(versions), len(pVersions))
 	toVersion := legacyLatestVersion + int64(postVersions)/2
-	err = tree.DeleteVersionsTo(toVersion)
+	err = tree.DeleteVersionsToSync(toVersion)
 	require.NoError(t, err)
 	pVersions = tree.AvailableVersions()
 	require.Equal(t, postVersions/2, len(pVersions))
 }
+
+func TestLazyPruning(t *testing.T) {
+	legacyVersion := 200
+	dbDir := fmt.Sprintf("./legacy-%s-%d", dbType, legacyVersion)
+	relateDir, err := createLegacyTree(t, dbDir, legacyVersion)
+	require.NoError(t, err)
+
+	db, err := dbm.NewDB("test", dbType, relateDir)
+	require.NoError(t, err)
+
+	defer func() {
+		if err := db.Close(); err != nil {
+			t.Errorf("DB close error: %v\n", err)
+		}
+		if err := os.RemoveAll(relateDir); err != nil {
+			t.Errorf("%+v\n", err)
+		}
+	}()
+
+	// Load the latest version
+	tree := NewMutableTree(db, 1000, false, log.NewNopLogger())
+	_, err = tree.Load()
+	require.NoError(t, err)
+
+	// Save 10 versions without updates
+	for i := 0; i < 10; i++ {
+		_, _, err = tree.SaveVersion()
+		require.NoError(t, err)
+	}
+
+	// Save 990 versions
+	leavesCount := 10
+	toVersion := int64(990)
+	pruningInterval := int64(20)
+	for i := int64(0); i < toVersion; i++ {
+		for j := 0; j < leavesCount; j++ {
+			_, err := tree.Set([]byte(fmt.Sprintf("key%d", j)), []byte(fmt.Sprintf("value%d", j)))
+			require.NoError(t, err)
+		}
+		_, v, err := tree.SaveVersion()
+		require.NoError(t, err)
+		if v%pruningInterval == 0 {
+			err = tree.DeleteVersionsTo(v - pruningInterval/2)
+			require.NoError(t, err)
+		}
+	}
+
+	// Wait for pruning to finish
+	for i := 0; i < 100; i++ {
+		_, _, err := tree.SaveVersion()
+		require.NoError(t, err)
+		firstVersion, err := tree.ndb.getFirstVersion()
+		require.NoError(t, err)
+		if firstVersion == toVersion+int64(legacyVersion)+1 {
+			break
+		}
+		// Simulate the consensus state update
+		time.Sleep(100 * time.Millisecond)
+	}
+
+	// Reload the tree
+	tree = NewMutableTree(db, 0, false, log.NewNopLogger())
+	versions := tree.AvailableVersions()
+	require.Equal(t, versions[0], int(toVersion)+legacyVersion+1)
+	for _, v := range versions {
+		_, err := tree.LoadVersion(int64(v))
+		require.NoError(t, err)
+	}
+	// Check if the legacy nodes are pruned
+	_, err = tree.Load()
+	require.NoError(t, err)
+	itr, err := NewNodeIterator(tree.root.GetKey(), tree.ndb)
+	require.NoError(t, err)
+	legacyNodes := make(map[string]*Node)
+	for ; itr.Valid(); itr.Next(false) {
+		node := itr.GetNode()
+		if node.nodeKey.nonce == 0 {
+			legacyNodes[string(node.hash)] = node
+		}
+	}
+
+	lNodes, err := tree.ndb.legacyNodes()
+	require.NoError(t, err)
+	require.Len(t, lNodes, len(legacyNodes))
+	for _, node := range lNodes {
+		_, ok := legacyNodes[string(node.hash)]
+		require.True(t, ok)
+	}
+}
+
+func TestRandomSet(t *testing.T) {
+	legacyVersion := 50
+	dbDir := fmt.Sprintf("./legacy-%s-%d", dbType, legacyVersion)
+	relateDir, err := createLegacyTree(t, dbDir, legacyVersion)
+	require.NoError(t, err)
+
+	db, err := dbm.NewDB("test", dbType, relateDir)
+	require.NoError(t, err)
+
+	defer func() {
+		if err := db.Close(); err != nil {
+			t.Errorf("DB close error: %v\n", err)
+		}
+		if err := os.RemoveAll(relateDir); err != nil {
+			t.Errorf("%+v\n", err)
+		}
+	}()
+
+	tree := NewMutableTree(db, 10000, false, log.NewNopLogger())
+
+	// Load the latest legacy version
+	_, err = tree.LoadVersion(int64(legacyVersion))
+	require.NoError(t, err)
+
+	// Commit new versions
+	postVersions := 1000
+	emptyVersions := 10
+	for i := 0; i < emptyVersions; i++ {
+		_, _, err := tree.SaveVersion()
+		require.NoError(t, err)
+	}
+	for i := 0; i < postVersions-emptyVersions; i++ {
+		leafCount := rand.Intn(50)
+		for j := 0; j < leafCount; j++ {
+			key := iavlrand.RandBytes(10)
+			value := iavlrand.RandBytes(10)
+			_, err = tree.Set(key, value)
+			require.NoError(t, err)
+		}
+		_, _, err = tree.SaveVersion()
+		require.NoError(t, err)
+	}
+
+	err = tree.DeleteVersionsToSync(int64(legacyVersion + postVersions - 1))
+	require.NoError(t, err)
+}