core/rawdb: allow for truncation in the freezer (ethereum#31362)

Here we add the notion of prunable tables for the `TruncateTail` operation
in the freezer. TruncateTail for the chain freezer now only truncates the body and
receipts tables, leaving headers and hashes as-is.

This change also requires changing the validation/repair at startup to allow for
tables with different tail. For the header and hash tables, we now require them to start
at number zero.

---------

Co-authored-by: Felix Lange <[email protected]>
Co-authored-by: Gary Rong <[email protected]>

Author: 3 people

core/rawdb/ancient_scheme.go

@@ -37,13 +37,21 @@ const (
 	ChainFreezerReceiptTable = "receipts"
 )
 
-// chainFreezerNoSnappy configures whether compression is disabled for the ancient-tables.
-// Hashes and difficulties don't compress well.
-var chainFreezerNoSnappy = map[string]bool{
-	ChainFreezerHeaderTable:  false,
-	ChainFreezerHashTable:    true,
-	ChainFreezerBodiesTable:  false,
-	ChainFreezerReceiptTable: false,
+// chainFreezerTableConfigs configures the settings for tables in the chain freezer.
+// Compression is disabled for hashes as they don't compress well. Additionally,
+// tail truncation is disabled for the header and hash tables, as these are intended
+// to be retained long-term.
+var chainFreezerTableConfigs = map[string]freezerTableConfig{
+	ChainFreezerHeaderTable:  {noSnappy: false, prunable: false},
+	ChainFreezerHashTable:    {noSnappy: true, prunable: false},
+	ChainFreezerBodiesTable:  {noSnappy: false, prunable: true},
+	ChainFreezerReceiptTable: {noSnappy: false, prunable: true},
+}
+
+// freezerTableConfig contains the settings for a freezer table.
+type freezerTableConfig struct {
+	noSnappy bool // disables item compression
+	prunable bool // true for tables that can be pruned by TruncateTail
 }
 
 const (
@@ -58,13 +66,13 @@ const (
 	stateHistoryStorageData  = "storage.data"
 )
 
-// stateFreezerNoSnappy configures whether compression is disabled for the state freezer.
-var stateFreezerNoSnappy = map[string]bool{
-	stateHistoryMeta:         true,
-	stateHistoryAccountIndex: false,
-	stateHistoryStorageIndex: false,
-	stateHistoryAccountData:  false,
-	stateHistoryStorageData:  false,
+// stateFreezerTableConfigs configures the settings for tables in the state freezer.
+var stateFreezerTableConfigs = map[string]freezerTableConfig{
+	stateHistoryMeta:         {noSnappy: true, prunable: true},
+	stateHistoryAccountIndex: {noSnappy: false, prunable: true},
+	stateHistoryStorageIndex: {noSnappy: false, prunable: true},
+	stateHistoryAccountData:  {noSnappy: false, prunable: true},
+	stateHistoryStorageData:  {noSnappy: false, prunable: true},
 }
 
 // The list of identifiers of ancient stores.
@@ -85,13 +93,13 @@ var freezers = []string{ChainFreezerName, MerkleStateFreezerName, VerkleStateFre
 //     state freezer.
 func NewStateFreezer(ancientDir string, verkle bool, readOnly bool) (ethdb.ResettableAncientStore, error) {
 	if ancientDir == "" {
-		return NewMemoryFreezer(readOnly, stateFreezerNoSnappy), nil
+		return NewMemoryFreezer(readOnly, stateFreezerTableConfigs), nil
 	}
 	var name string
 	if verkle {
 		name = filepath.Join(ancientDir, VerkleStateFreezerName)
 	} else {
 		name = filepath.Join(ancientDir, MerkleStateFreezerName)
 	}
-	return newResettableFreezer(name, "eth/db/state", readOnly, stateHistoryTableSize, stateFreezerNoSnappy)
+	return newResettableFreezer(name, "eth/db/state", readOnly, stateHistoryTableSize, stateFreezerTableConfigs)
 }

core/rawdb/ancient_utils.go

@@ -51,7 +51,7 @@ func (info *freezerInfo) size() common.StorageSize {
 	return total
 }
 
-func inspect(name string, order map[string]bool, reader ethdb.AncientReader) (freezerInfo, error) {
+func inspect(name string, order map[string]freezerTableConfig, reader ethdb.AncientReader) (freezerInfo, error) {
 	info := freezerInfo{name: name}
 	for t := range order {
 		size, err := reader.AncientSize(t)
@@ -82,7 +82,7 @@ func inspectFreezers(db ethdb.Database) ([]freezerInfo, error) {
 	for _, freezer := range freezers {
 		switch freezer {
 		case ChainFreezerName:
-			info, err := inspect(ChainFreezerName, chainFreezerNoSnappy, db)
+			info, err := inspect(ChainFreezerName, chainFreezerTableConfigs, db)
 			if err != nil {
 				return nil, err
 			}
@@ -99,7 +99,7 @@ func inspectFreezers(db ethdb.Database) ([]freezerInfo, error) {
 			}
 			defer f.Close()
 
-			info, err := inspect(freezer, stateFreezerNoSnappy, f)
+			info, err := inspect(freezer, stateFreezerTableConfigs, f)
 			if err != nil {
 				return nil, err
 			}
@@ -119,13 +119,13 @@ func inspectFreezers(db ethdb.Database) ([]freezerInfo, error) {
 func InspectFreezerTable(ancient string, freezerName string, tableName string, start, end int64) error {
 	var (
 		path   string
-		tables map[string]bool
+		tables map[string]freezerTableConfig
 	)
 	switch freezerName {
 	case ChainFreezerName:
-		path, tables = resolveChainFreezerDir(ancient), chainFreezerNoSnappy
+		path, tables = resolveChainFreezerDir(ancient), chainFreezerTableConfigs
 	case MerkleStateFreezerName, VerkleStateFreezerName:
-		path, tables = filepath.Join(ancient, freezerName), stateFreezerNoSnappy
+		path, tables = filepath.Join(ancient, freezerName), stateFreezerTableConfigs
 	default:
 		return fmt.Errorf("unknown freezer, supported ones: %v", freezers)
 	}

core/rawdb/chain_freezer.go

@@ -62,9 +62,9 @@ func newChainFreezer(datadir string, namespace string, readonly bool) (*chainFre
 		freezer ethdb.AncientStore
 	)
 	if datadir == "" {
-		freezer = NewMemoryFreezer(readonly, chainFreezerNoSnappy)
+		freezer = NewMemoryFreezer(readonly, chainFreezerTableConfigs)
 	} else {
-		freezer, err = NewFreezer(datadir, namespace, readonly, freezerTableSize, chainFreezerNoSnappy)
+		freezer, err = NewFreezer(datadir, namespace, readonly, freezerTableSize, chainFreezerTableConfigs)
 	}
 	if err != nil {
 		return nil, err

core/rawdb/freezer.go

@@ -78,7 +78,7 @@ type Freezer struct {
 //
 // The 'tables' argument defines the data tables. If the value of a map
 // entry is true, snappy compression is disabled for the table.
-func NewFreezer(datadir string, namespace string, readonly bool, maxTableSize uint32, tables map[string]bool) (*Freezer, error) {
+func NewFreezer(datadir string, namespace string, readonly bool, maxTableSize uint32, tables map[string]freezerTableConfig) (*Freezer, error) {
 	// Create the initial freezer object
 	var (
 		readMeter  = metrics.NewRegisteredMeter(namespace+"ancient/read", nil)
@@ -121,8 +121,8 @@ func NewFreezer(datadir string, namespace string, readonly bool, maxTableSize ui
 	}
 
 	// Create the tables.
-	for name, disableSnappy := range tables {
-		table, err := newTable(datadir, name, readMeter, writeMeter, sizeGauge, maxTableSize, disableSnappy, readonly)
+	for name, config := range tables {
+		table, err := newTable(datadir, name, readMeter, writeMeter, sizeGauge, maxTableSize, config, readonly)
 		if err != nil {
 			for _, table := range freezer.tables {
 				table.Close()
@@ -301,7 +301,8 @@ func (f *Freezer) TruncateHead(items uint64) (uint64, error) {
 	return oitems, nil
 }
 
-// TruncateTail discards any recent data below the provided threshold number.
+// TruncateTail discards all data below the specified threshold. Note that only
+// 'prunable' tables will be truncated.
 func (f *Freezer) TruncateTail(tail uint64) (uint64, error) {
 	if f.readonly {
 		return 0, errReadOnly
@@ -314,8 +315,10 @@ func (f *Freezer) TruncateTail(tail uint64) (uint64, error) {
 		return old, nil
 	}
 	for _, table := range f.tables {
-		if err := table.truncateTail(tail); err != nil {
-			return 0, err
+		if table.config.prunable {
+			if err := table.truncateTail(tail); err != nil {
+				return 0, err
+			}
 		}
 	}
 	f.tail.Store(tail)
@@ -343,56 +346,77 @@ func (f *Freezer) validate() error {
 		return nil
 	}
 	var (
-		head uint64
-		tail uint64
-		name string
+		head       uint64
+		prunedTail *uint64
 	)
-	// Hack to get boundary of any table
-	for kind, table := range f.tables {
+	// get any head value
+	for _, table := range f.tables {
 		head = table.items.Load()
-		tail = table.itemHidden.Load()
-		name = kind
 		break
 	}
-	// Now check every table against those boundaries.
 	for kind, table := range f.tables {
+		// all tables have to have the same head
 		if head != table.items.Load() {
-			return fmt.Errorf("freezer tables %s and %s have differing head: %d != %d", kind, name, table.items.Load(), head)
+			return fmt.Errorf("freezer table %s has a differing head: %d != %d", kind, table.items.Load(), head)
 		}
-		if tail != table.itemHidden.Load() {
-			return fmt.Errorf("freezer tables %s and %s have differing tail: %d != %d", kind, name, table.itemHidden.Load(), tail)
+		if !table.config.prunable {
+			// non-prunable tables have to start at 0
+			if table.itemHidden.Load() != 0 {
+				return fmt.Errorf("non-prunable freezer table '%s' has a non-zero tail: %d", kind, table.itemHidden.Load())
+			}
+		} else {
+			// prunable tables have to have the same length
+			if prunedTail == nil {
+				tmp := table.itemHidden.Load()
+				prunedTail = &tmp
+			}
+			if *prunedTail != table.itemHidden.Load() {
+				return fmt.Errorf("freezer table %s has differing tail: %d != %d", kind, table.itemHidden.Load(), *prunedTail)
+			}
 		}
 	}
+
+	if prunedTail == nil {
+		tmp := uint64(0)
+		prunedTail = &tmp
+	}
+
 	f.frozen.Store(head)
-	f.tail.Store(tail)
+	f.tail.Store(*prunedTail)
 	return nil
 }
 
 // repair truncates all data tables to the same length.
 func (f *Freezer) repair() error {
 	var (
-		head = uint64(math.MaxUint64)
-		tail = uint64(0)
+		head       = uint64(math.MaxUint64)
+		prunedTail = uint64(0)
 	)
+	// get the minimal head and the maximum tail
 	for _, table := range f.tables {
-		items := table.items.Load()
-		if head > items {
-			head = items
-		}
-		hidden := table.itemHidden.Load()
-		if hidden > tail {
-			tail = hidden
-		}
+		head = min(head, table.items.Load())
+		prunedTail = max(prunedTail, table.itemHidden.Load())
 	}
-	for _, table := range f.tables {
+	// apply the pruning
+	for kind, table := range f.tables {
+		// all tables need to have the same head
 		if err := table.truncateHead(head); err != nil {
 			return err
 		}
-		if err := table.truncateTail(tail); err != nil {
-			return err
+		if !table.config.prunable {
+			// non-prunable tables have to start at 0
+			if table.itemHidden.Load() != 0 {
+				panic(fmt.Sprintf("non-prunable freezer table %s has non-zero tail: %v", kind, table.itemHidden.Load()))
+			}
+		} else {
+			// prunable tables have to have the same length
+			if err := table.truncateTail(prunedTail); err != nil {
+				return err
+			}
 		}
 	}
+
 	f.frozen.Store(head)
-	f.tail.Store(tail)
+	f.tail.Store(prunedTail)
 	return nil
 }

core/rawdb/freezer_batch.go

@@ -96,7 +96,7 @@ type freezerTableBatch struct {
 // newBatch creates a new batch for the freezer table.
 func (t *freezerTable) newBatch() *freezerTableBatch {
 	batch := &freezerTableBatch{t: t}
-	if !t.noCompression {
+	if !t.config.noSnappy {
 		batch.sb = new(snappyBuffer)
 	}
 	batch.reset()

core/rawdb/freezer_memory.go

@@ -30,17 +30,19 @@ import (
 
 // memoryTable is used to store a list of sequential items in memory.
 type memoryTable struct {
-	name   string   // Table name
 	items  uint64   // Number of stored items in the table, including the deleted ones
 	offset uint64   // Number of deleted items from the table
 	data   [][]byte // List of rlp-encoded items, sort in order
 	size   uint64   // Total memory size occupied by the table
 	lock   sync.RWMutex
+
+	name   string
+	config freezerTableConfig
 }
 
 // newMemoryTable initializes the memory table.
-func newMemoryTable(name string) *memoryTable {
-	return &memoryTable{name: name}
+func newMemoryTable(name string, config freezerTableConfig) *memoryTable {
+	return &memoryTable{name: name, config: config}
 }
 
 // has returns an indicator whether the specified data exists.
@@ -218,10 +220,10 @@ type MemoryFreezer struct {
 }
 
 // NewMemoryFreezer initializes an in-memory freezer instance.
-func NewMemoryFreezer(readonly bool, tableName map[string]bool) *MemoryFreezer {
+func NewMemoryFreezer(readonly bool, tableName map[string]freezerTableConfig) *MemoryFreezer {
 	tables := make(map[string]*memoryTable)
-	for name := range tableName {
-		tables[name] = newMemoryTable(name)
+	for name, cfg := range tableName {
+		tables[name] = newMemoryTable(name, cfg)
 	}
 	return &MemoryFreezer{
 		writeBatch: newMemoryBatch(),
@@ -368,7 +370,9 @@ func (f *MemoryFreezer) TruncateHead(items uint64) (uint64, error) {
 	return old, nil
 }
 
-// TruncateTail discards any recent data below the provided threshold number.
+// TruncateTail discards all data below the provided threshold number.
+// Note this will only truncate 'prunable' tables. Block headers and canonical
+// hashes cannot be truncated at this time.
 func (f *MemoryFreezer) TruncateTail(tail uint64) (uint64, error) {
 	f.lock.Lock()
 	defer f.lock.Unlock()
@@ -381,8 +385,10 @@ func (f *MemoryFreezer) TruncateTail(tail uint64) (uint64, error) {
 		return old, nil
 	}
 	for _, table := range f.tables {
-		if err := table.truncateTail(tail); err != nil {
-			return 0, err
+		if table.config.prunable {
+			if err := table.truncateTail(tail); err != nil {
+				return 0, err
+			}
 		}
 	}
 	f.tail = tail
@@ -412,8 +418,8 @@ func (f *MemoryFreezer) Reset() error {
 	defer f.lock.Unlock()
 
 	tables := make(map[string]*memoryTable)
-	for name := range f.tables {
-		tables[name] = newMemoryTable(name)
+	for name, table := range f.tables {
+		tables[name] = newMemoryTable(name, table.config)
 	}
 	f.tables = tables
 	f.items, f.tail = 0, 0

core/rawdb/freezer_memory_test.go

@@ -25,16 +25,22 @@ import (
 
 func TestMemoryFreezer(t *testing.T) {
 	ancienttest.TestAncientSuite(t, func(kinds []string) ethdb.AncientStore {
-		tables := make(map[string]bool)
+		tables := make(map[string]freezerTableConfig)
 		for _, kind := range kinds {
-			tables[kind] = true
+			tables[kind] = freezerTableConfig{
+				noSnappy: true,
+				prunable: true,
+			}
 		}
 		return NewMemoryFreezer(false, tables)
 	})
 	ancienttest.TestResettableAncientSuite(t, func(kinds []string) ethdb.ResettableAncientStore {
-		tables := make(map[string]bool)
+		tables := make(map[string]freezerTableConfig)
 		for _, kind := range kinds {
-			tables[kind] = true
+			tables[kind] = freezerTableConfig{
+				noSnappy: true,
+				prunable: true,
+			}
 		}
 		return NewMemoryFreezer(false, tables)
 	})

core/rawdb/freezer_resettable.go

@@ -49,7 +49,7 @@ type resettableFreezer struct {
 //
 // The reset function will delete directory atomically and re-create the
 // freezer from scratch.
-func newResettableFreezer(datadir string, namespace string, readonly bool, maxTableSize uint32, tables map[string]bool) (*resettableFreezer, error) {
+func newResettableFreezer(datadir string, namespace string, readonly bool, maxTableSize uint32, tables map[string]freezerTableConfig) (*resettableFreezer, error) {
 	if err := cleanup(datadir); err != nil {
 		return nil, err
 	}