Official Go implementation of the Ethereum protocol
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
go-ethereum/core/rawdb/database.go

685 lines
26 KiB

// Copyright 2018 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package rawdb
import (
"bytes"
"errors"
"fmt"
"os"
"path/filepath"
"strings"
"time"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/ethdb"
"github.com/ethereum/go-ethereum/ethdb/leveldb"
"github.com/ethereum/go-ethereum/ethdb/memorydb"
"github.com/ethereum/go-ethereum/ethdb/pebble"
"github.com/ethereum/go-ethereum/log"
"github.com/olekukonko/tablewriter"
)
// freezerdb is a database wrapper that enables ancient chain segment freezing.
type freezerdb struct {
ethdb.KeyValueStore
*chainFreezer
readOnly bool
ancientRoot string
}
// AncientDatadir returns the path of root ancient directory.
func (frdb *freezerdb) AncientDatadir() (string, error) {
return frdb.ancientRoot, nil
}
// Close implements io.Closer, closing both the fast key-value store as well as
// the slow ancient tables.
func (frdb *freezerdb) Close() error {
var errs []error
if err := frdb.chainFreezer.Close(); err != nil {
errs = append(errs, err)
}
if err := frdb.KeyValueStore.Close(); err != nil {
errs = append(errs, err)
}
if len(errs) != 0 {
return fmt.Errorf("%v", errs)
}
return nil
}
// Freeze is a helper method used for external testing to trigger and block until
// a freeze cycle completes, without having to sleep for a minute to trigger the
// automatic background run.
func (frdb *freezerdb) Freeze() error {
if frdb.readOnly {
return errReadOnly
}
// Trigger a freeze cycle and block until it's done
trigger := make(chan struct{}, 1)
frdb.chainFreezer.trigger <- trigger
<-trigger
return nil
}
// nofreezedb is a database wrapper that disables freezer data retrievals.
type nofreezedb struct {
ethdb.KeyValueStore
}
// HasAncient returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) HasAncient(kind string, number uint64) (bool, error) {
return false, errNotSupported
}
// Ancient returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) Ancient(kind string, number uint64) ([]byte, error) {
return nil, errNotSupported
}
// AncientRange returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) AncientRange(kind string, start, max, maxByteSize uint64) ([][]byte, error) {
return nil, errNotSupported
}
// Ancients returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) Ancients() (uint64, error) {
return 0, errNotSupported
}
// Tail returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) Tail() (uint64, error) {
return 0, errNotSupported
}
// AncientSize returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) AncientSize(kind string) (uint64, error) {
return 0, errNotSupported
}
// ModifyAncients is not supported.
func (db *nofreezedb) ModifyAncients(func(ethdb.AncientWriteOp) error) (int64, error) {
return 0, errNotSupported
}
// TruncateHead returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) TruncateHead(items uint64) (uint64, error) {
return 0, errNotSupported
}
// TruncateTail returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) TruncateTail(items uint64) (uint64, error) {
return 0, errNotSupported
}
// Sync returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) Sync() error {
return errNotSupported
}
func (db *nofreezedb) ReadAncients(fn func(reader ethdb.AncientReaderOp) error) (err error) {
// Unlike other ancient-related methods, this method does not return
// errNotSupported when invoked.
// The reason for this is that the caller might want to do several things:
// 1. Check if something is in the freezer,
// 2. If not, check leveldb.
//
// This will work, since the ancient-checks inside 'fn' will return errors,
// and the leveldb work will continue.
//
// If we instead were to return errNotSupported here, then the caller would
// have to explicitly check for that, having an extra clause to do the
// non-ancient operations.
return fn(db)
}
// AncientDatadir returns an error as we don't have a backing chain freezer.
func (db *nofreezedb) AncientDatadir() (string, error) {
return "", errNotSupported
}
// NewDatabase creates a high level database on top of a given key-value data
// store without a freezer moving immutable chain segments into cold storage.
func NewDatabase(db ethdb.KeyValueStore) ethdb.Database {
return &nofreezedb{KeyValueStore: db}
}
// resolveChainFreezerDir is a helper function which resolves the absolute path
// of chain freezer by considering backward compatibility.
func resolveChainFreezerDir(ancient string) string {
// Check if the chain freezer is already present in the specified
// sub folder, if not then two possibilities:
// - chain freezer is not initialized
// - chain freezer exists in legacy location (root ancient folder)
freezer := filepath.Join(ancient, ChainFreezerName)
if !common.FileExist(freezer) {
if !common.FileExist(ancient) {
// The entire ancient store is not initialized, still use the sub
// folder for initialization.
} else {
// Ancient root is already initialized, then we hold the assumption
// that chain freezer is also initialized and located in root folder.
// In this case fallback to legacy location.
freezer = ancient
log.Info("Found legacy ancient chain path", "location", ancient)
}
}
return freezer
}
// NewDatabaseWithFreezer creates a high level database on top of a given key-
// value data store with a freezer moving immutable chain segments into cold
// storage. The passed ancient indicates the path of root ancient directory
// where the chain freezer can be opened.
func NewDatabaseWithFreezer(db ethdb.KeyValueStore, ancient string, namespace string, readonly bool) (ethdb.Database, error) {
// Create the idle freezer instance. If the given ancient directory is empty,
// in-memory chain freezer is used (e.g. dev mode); otherwise the regular
// file-based freezer is created.
chainFreezerDir := ancient
if chainFreezerDir != "" {
chainFreezerDir = resolveChainFreezerDir(chainFreezerDir)
}
frdb, err := newChainFreezer(chainFreezerDir, namespace, readonly)
if err != nil {
printChainMetadata(db)
return nil, err
}
// Since the freezer can be stored separately from the user's key-value database,
// there's a fairly high probability that the user requests invalid combinations
// of the freezer and database. Ensure that we don't shoot ourselves in the foot
// by serving up conflicting data, leading to both datastores getting corrupted.
//
// - If both the freezer and key-value store are empty (no genesis), we just
// initialized a new empty freezer, so everything's fine.
// - If the key-value store is empty, but the freezer is not, we need to make
// sure the user's genesis matches the freezer. That will be checked in the
// blockchain, since we don't have the genesis block here (nor should we at
// this point care, the key-value/freezer combo is valid).
// - If neither the key-value store nor the freezer is empty, cross validate
// the genesis hashes to make sure they are compatible. If they are, also
// ensure that there's no gap between the freezer and subsequently leveldb.
// - If the key-value store is not empty, but the freezer is, we might just be
// upgrading to the freezer release, or we might have had a small chain and
// not frozen anything yet. Ensure that no blocks are missing yet from the
// key-value store, since that would mean we already had an old freezer.
// If the genesis hash is empty, we have a new key-value store, so nothing to
// validate in this method. If, however, the genesis hash is not nil, compare
// it to the freezer content.
if kvgenesis, _ := db.Get(headerHashKey(0)); len(kvgenesis) > 0 {
if frozen, _ := frdb.Ancients(); frozen > 0 {
// If the freezer already contains something, ensure that the genesis blocks
// match, otherwise we might mix up freezers across chains and destroy both
// the freezer and the key-value store.
frgenesis, err := frdb.Ancient(ChainFreezerHashTable, 0)
if err != nil {
printChainMetadata(db)
return nil, fmt.Errorf("failed to retrieve genesis from ancient %v", err)
} else if !bytes.Equal(kvgenesis, frgenesis) {
printChainMetadata(db)
return nil, fmt.Errorf("genesis mismatch: %#x (leveldb) != %#x (ancients)", kvgenesis, frgenesis)
}
// Key-value store and freezer belong to the same network. Ensure that they
// are contiguous, otherwise we might end up with a non-functional freezer.
if kvhash, _ := db.Get(headerHashKey(frozen)); len(kvhash) == 0 {
// Subsequent header after the freezer limit is missing from the database.
// Reject startup if the database has a more recent head.
if head := *ReadHeaderNumber(db, ReadHeadHeaderHash(db)); head > frozen-1 {
// Find the smallest block stored in the key-value store
// in range of [frozen, head]
var number uint64
for number = frozen; number <= head; number++ {
if present, _ := db.Has(headerHashKey(number)); present {
break
}
}
// We are about to exit on error. Print database metadata before exiting
printChainMetadata(db)
return nil, fmt.Errorf("gap in the chain between ancients [0 - #%d] and leveldb [#%d - #%d] ",
frozen-1, number, head)
}
// Database contains only older data than the freezer, this happens if the
// state was wiped and reinited from an existing freezer.
}
// Otherwise, key-value store continues where the freezer left off, all is fine.
// We might have duplicate blocks (crash after freezer write but before key-value
// store deletion, but that's fine).
} else {
// If the freezer is empty, ensure nothing was moved yet from the key-value
// store, otherwise we'll end up missing data. We check block #1 to decide
// if we froze anything previously or not, but do take care of databases with
// only the genesis block.
if ReadHeadHeaderHash(db) != common.BytesToHash(kvgenesis) {
// Key-value store contains more data than the genesis block, make sure we
// didn't freeze anything yet.
if kvblob, _ := db.Get(headerHashKey(1)); len(kvblob) == 0 {
printChainMetadata(db)
return nil, errors.New("ancient chain segments already extracted, please set --datadir.ancient to the correct path")
}
// Block #1 is still in the database, we're allowed to init a new freezer
}
// Otherwise, the head header is still the genesis, we're allowed to init a new
// freezer.
}
}
// Freezer is consistent with the key-value database, permit combining the two
if !readonly {
frdb.wg.Add(1)
go func() {
frdb.freeze(db)
frdb.wg.Done()
}()
}
return &freezerdb{
ancientRoot: ancient,
KeyValueStore: db,
chainFreezer: frdb,
}, nil
}
// NewMemoryDatabase creates an ephemeral in-memory key-value database without a
// freezer moving immutable chain segments into cold storage.
func NewMemoryDatabase() ethdb.Database {
return NewDatabase(memorydb.New())
}
// NewMemoryDatabaseWithCap creates an ephemeral in-memory key-value database
// with an initial starting capacity, but without a freezer moving immutable
// chain segments into cold storage.
func NewMemoryDatabaseWithCap(size int) ethdb.Database {
return NewDatabase(memorydb.NewWithCap(size))
}
// NewLevelDBDatabase creates a persistent key-value database without a freezer
// moving immutable chain segments into cold storage.
func NewLevelDBDatabase(file string, cache int, handles int, namespace string, readonly bool) (ethdb.Database, error) {
db, err := leveldb.New(file, cache, handles, namespace, readonly)
if err != nil {
return nil, err
}
log.Info("Using LevelDB as the backing database")
return NewDatabase(db), nil
}
// NewPebbleDBDatabase creates a persistent key-value database without a freezer
// moving immutable chain segments into cold storage.
func NewPebbleDBDatabase(file string, cache int, handles int, namespace string, readonly, ephemeral bool) (ethdb.Database, error) {
db, err := pebble.New(file, cache, handles, namespace, readonly, ephemeral)
if err != nil {
return nil, err
}
return NewDatabase(db), nil
}
const (
dbPebble = "pebble"
dbLeveldb = "leveldb"
)
// PreexistingDatabase checks the given data directory whether a database is already
// instantiated at that location, and if so, returns the type of database (or the
// empty string).
func PreexistingDatabase(path string) string {
if _, err := os.Stat(filepath.Join(path, "CURRENT")); err != nil {
return "" // No pre-existing db
}
if matches, err := filepath.Glob(filepath.Join(path, "OPTIONS*")); len(matches) > 0 || err != nil {
if err != nil {
panic(err) // only possible if the pattern is malformed
}
return dbPebble
}
return dbLeveldb
}
// OpenOptions contains the options to apply when opening a database.
// OBS: If AncientsDirectory is empty, it indicates that no freezer is to be used.
type OpenOptions struct {
Type string // "leveldb" | "pebble"
Directory string // the datadir
AncientsDirectory string // the ancients-dir
Namespace string // the namespace for database relevant metrics
Cache int // the capacity(in megabytes) of the data caching
Handles int // number of files to be open simultaneously
ReadOnly bool
// Ephemeral means that filesystem sync operations should be avoided: data integrity in the face of
// a crash is not important. This option should typically be used in tests.
Ephemeral bool
}
// openKeyValueDatabase opens a disk-based key-value database, e.g. leveldb or pebble.
//
// type == null type != null
// +----------------------------------------
// db is non-existent | pebble default | specified type
// db is existent | from db | specified type (if compatible)
func openKeyValueDatabase(o OpenOptions) (ethdb.Database, error) {
// Reject any unsupported database type
if len(o.Type) != 0 && o.Type != dbLeveldb && o.Type != dbPebble {
return nil, fmt.Errorf("unknown db.engine %v", o.Type)
}
// Retrieve any pre-existing database's type and use that or the requested one
// as long as there's no conflict between the two types
existingDb := PreexistingDatabase(o.Directory)
if len(existingDb) != 0 && len(o.Type) != 0 && o.Type != existingDb {
return nil, fmt.Errorf("db.engine choice was %v but found pre-existing %v database in specified data directory", o.Type, existingDb)
}
if o.Type == dbPebble || existingDb == dbPebble {
log.Info("Using pebble as the backing database")
return NewPebbleDBDatabase(o.Directory, o.Cache, o.Handles, o.Namespace, o.ReadOnly, o.Ephemeral)
}
if o.Type == dbLeveldb || existingDb == dbLeveldb {
log.Info("Using leveldb as the backing database")
return NewLevelDBDatabase(o.Directory, o.Cache, o.Handles, o.Namespace, o.ReadOnly)
}
// No pre-existing database, no user-requested one either. Default to Pebble.
log.Info("Defaulting to pebble as the backing database")
return NewPebbleDBDatabase(o.Directory, o.Cache, o.Handles, o.Namespace, o.ReadOnly, o.Ephemeral)
}
// Open opens both a disk-based key-value database such as leveldb or pebble, but also
// integrates it with a freezer database -- if the AncientDir option has been
// set on the provided OpenOptions.
// The passed o.AncientDir indicates the path of root ancient directory where
// the chain freezer can be opened.
func Open(o OpenOptions) (ethdb.Database, error) {
kvdb, err := openKeyValueDatabase(o)
if err != nil {
return nil, err
}
if len(o.AncientsDirectory) == 0 {
return kvdb, nil
}
frdb, err := NewDatabaseWithFreezer(kvdb, o.AncientsDirectory, o.Namespace, o.ReadOnly)
if err != nil {
kvdb.Close()
return nil, err
}
return frdb, nil
}
type counter uint64
func (c counter) String() string {
return fmt.Sprintf("%d", c)
}
func (c counter) Percentage(current uint64) string {
return fmt.Sprintf("%d", current*100/uint64(c))
}
// stat stores sizes and count for a parameter
type stat struct {
size common.StorageSize
count counter
}
// Add size to the stat and increase the counter by 1
func (s *stat) Add(size common.StorageSize) {
s.size += size
s.count++
}
func (s *stat) Size() string {
return s.size.String()
}
func (s *stat) Count() string {
return s.count.String()
}
// InspectDatabase traverses the entire database and checks the size
// of all different categories of data.
func InspectDatabase(db ethdb.Database, keyPrefix, keyStart []byte) error {
it := db.NewIterator(keyPrefix, keyStart)
defer it.Release()
var (
count int64
start = time.Now()
logged = time.Now()
// Key-value store statistics
headers stat
bodies stat
receipts stat
tds stat
numHashPairings stat
hashNumPairings stat
legacyTries stat
stateLookups stat
accountTries stat
storageTries stat
codes stat
txLookups stat
accountSnaps stat
storageSnaps stat
preimages stat
bloomBits stat
beaconHeaders stat
cliqueSnaps stat
// Verkle statistics
verkleTries stat
verkleStateLookups stat
// Les statistic
chtTrieNodes stat
bloomTrieNodes stat
// Meta- and unaccounted data
metadata stat
unaccounted stat
// Totals
total common.StorageSize
)
// Inspect key-value database first.
for it.Next() {
var (
key = it.Key()
size = common.StorageSize(len(key) + len(it.Value()))
)
total += size
switch {
case bytes.HasPrefix(key, headerPrefix) && len(key) == (len(headerPrefix)+8+common.HashLength):
headers.Add(size)
case bytes.HasPrefix(key, blockBodyPrefix) && len(key) == (len(blockBodyPrefix)+8+common.HashLength):
bodies.Add(size)
case bytes.HasPrefix(key, blockReceiptsPrefix) && len(key) == (len(blockReceiptsPrefix)+8+common.HashLength):
receipts.Add(size)
case bytes.HasPrefix(key, headerPrefix) && bytes.HasSuffix(key, headerTDSuffix):
tds.Add(size)
case bytes.HasPrefix(key, headerPrefix) && bytes.HasSuffix(key, headerHashSuffix):
numHashPairings.Add(size)
case bytes.HasPrefix(key, headerNumberPrefix) && len(key) == (len(headerNumberPrefix)+common.HashLength):
hashNumPairings.Add(size)
case IsLegacyTrieNode(key, it.Value()):
legacyTries.Add(size)
case bytes.HasPrefix(key, stateIDPrefix) && len(key) == len(stateIDPrefix)+common.HashLength:
stateLookups.Add(size)
case IsAccountTrieNode(key):
accountTries.Add(size)
case IsStorageTrieNode(key):
storageTries.Add(size)
all: bloom-filter based pruning mechanism (#21724) * cmd, core, tests: initial state pruner core: fix db inspector cmd/geth: add verify-state cmd/geth: add verification tool core/rawdb: implement flatdb cmd, core: fix rebase core/state: use new contract code layout core/state/pruner: avoid deleting genesis state cmd/geth: add helper function core, cmd: fix extract genesis core: minor fixes contracts: remove useless core/state/snapshot: plugin stacktrie core: polish core/state/snapshot: iterate storage concurrently core/state/snapshot: fix iteration core: add comments core/state/snapshot: polish code core/state: polish core/state/snapshot: rebase core/rawdb: add comments core/rawdb: fix tests core/rawdb: improve tests core/state/snapshot: fix concurrent iteration core/state: run pruning during the recovery core, trie: implement martin's idea core, eth: delete flatdb and polish pruner trie: fix import core/state/pruner: add log core/state/pruner: fix issues core/state/pruner: don't read back core/state/pruner: fix contract code write core/state/pruner: check root node presence cmd, core: polish log core/state: use HEAD-127 as the target core/state/snapshot: improve tests cmd/geth: fix verification tool cmd/geth: use HEAD as the verification default target all: replace the bloomfilter with martin's fork cmd, core: polish code core, cmd: forcibly delete state root core/state/pruner: add hash64 core/state/pruner: fix blacklist core/state: remove blacklist cmd, core: delete trie clean cache before pruning cmd, core: fix lint cmd, core: fix rebase core/state: fix the special case for clique networks core/state/snapshot: remove useless code core/state/pruner: capping the snapshot after pruning cmd, core, eth: fixes core/rawdb: update db inspector cmd/geth: polish code core/state/pruner: fsync bloom filter cmd, core: print warning log core/state/pruner: adjust the parameters for bloom filter cmd, core: create the bloom filter by size core: polish core/state/pruner: sanitize invalid bloomfilter size cmd: address comments cmd/geth: address comments cmd/geth: address comment core/state/pruner: address comments core/state/pruner: rename homedir to datadir cmd, core: address comments core/state/pruner: address comment core/state: address comments core, cmd, tests: address comments core: address comments core/state/pruner: release the iterator after each commit core/state/pruner: improve pruner cmd, core: adjust bloom paramters core/state/pruner: fix lint core/state/pruner: fix tests core: fix rebase core/state/pruner: remove atomic rename core/state/pruner: address comments all: run go mod tidy core/state/pruner: avoid false-positive for the middle state roots core/state/pruner: add checks for middle roots cmd/geth: replace crit with error * core/state/pruner: fix lint * core: drop legacy bloom filter * core/state/snapshot: improve pruner * core/state/snapshot: polish concurrent logs to report ETA vs. hashes * core/state/pruner: add progress report for pruning and compaction too * core: fix snapshot test API * core/state: fix some pruning logs * core/state/pruner: support recovering from bloom flush fail Co-authored-by: Péter Szilágyi <peterke@gmail.com>
4 years ago
case bytes.HasPrefix(key, CodePrefix) && len(key) == len(CodePrefix)+common.HashLength:
codes.Add(size)
case bytes.HasPrefix(key, txLookupPrefix) && len(key) == (len(txLookupPrefix)+common.HashLength):
txLookups.Add(size)
case bytes.HasPrefix(key, SnapshotAccountPrefix) && len(key) == (len(SnapshotAccountPrefix)+common.HashLength):
accountSnaps.Add(size)
case bytes.HasPrefix(key, SnapshotStoragePrefix) && len(key) == (len(SnapshotStoragePrefix)+2*common.HashLength):
storageSnaps.Add(size)
case bytes.HasPrefix(key, PreimagePrefix) && len(key) == (len(PreimagePrefix)+common.HashLength):
preimages.Add(size)
case bytes.HasPrefix(key, configPrefix) && len(key) == (len(configPrefix)+common.HashLength):
metadata.Add(size)
case bytes.HasPrefix(key, genesisPrefix) && len(key) == (len(genesisPrefix)+common.HashLength):
metadata.Add(size)
case bytes.HasPrefix(key, bloomBitsPrefix) && len(key) == (len(bloomBitsPrefix)+10+common.HashLength):
bloomBits.Add(size)
all: bloom-filter based pruning mechanism (#21724) * cmd, core, tests: initial state pruner core: fix db inspector cmd/geth: add verify-state cmd/geth: add verification tool core/rawdb: implement flatdb cmd, core: fix rebase core/state: use new contract code layout core/state/pruner: avoid deleting genesis state cmd/geth: add helper function core, cmd: fix extract genesis core: minor fixes contracts: remove useless core/state/snapshot: plugin stacktrie core: polish core/state/snapshot: iterate storage concurrently core/state/snapshot: fix iteration core: add comments core/state/snapshot: polish code core/state: polish core/state/snapshot: rebase core/rawdb: add comments core/rawdb: fix tests core/rawdb: improve tests core/state/snapshot: fix concurrent iteration core/state: run pruning during the recovery core, trie: implement martin's idea core, eth: delete flatdb and polish pruner trie: fix import core/state/pruner: add log core/state/pruner: fix issues core/state/pruner: don't read back core/state/pruner: fix contract code write core/state/pruner: check root node presence cmd, core: polish log core/state: use HEAD-127 as the target core/state/snapshot: improve tests cmd/geth: fix verification tool cmd/geth: use HEAD as the verification default target all: replace the bloomfilter with martin's fork cmd, core: polish code core, cmd: forcibly delete state root core/state/pruner: add hash64 core/state/pruner: fix blacklist core/state: remove blacklist cmd, core: delete trie clean cache before pruning cmd, core: fix lint cmd, core: fix rebase core/state: fix the special case for clique networks core/state/snapshot: remove useless code core/state/pruner: capping the snapshot after pruning cmd, core, eth: fixes core/rawdb: update db inspector cmd/geth: polish code core/state/pruner: fsync bloom filter cmd, core: print warning log core/state/pruner: adjust the parameters for bloom filter cmd, core: create the bloom filter by size core: polish core/state/pruner: sanitize invalid bloomfilter size cmd: address comments cmd/geth: address comments cmd/geth: address comment core/state/pruner: address comments core/state/pruner: rename homedir to datadir cmd, core: address comments core/state/pruner: address comment core/state: address comments core, cmd, tests: address comments core: address comments core/state/pruner: release the iterator after each commit core/state/pruner: improve pruner cmd, core: adjust bloom paramters core/state/pruner: fix lint core/state/pruner: fix tests core: fix rebase core/state/pruner: remove atomic rename core/state/pruner: address comments all: run go mod tidy core/state/pruner: avoid false-positive for the middle state roots core/state/pruner: add checks for middle roots cmd/geth: replace crit with error * core/state/pruner: fix lint * core: drop legacy bloom filter * core/state/snapshot: improve pruner * core/state/snapshot: polish concurrent logs to report ETA vs. hashes * core/state/pruner: add progress report for pruning and compaction too * core: fix snapshot test API * core/state: fix some pruning logs * core/state/pruner: support recovering from bloom flush fail Co-authored-by: Péter Szilágyi <peterke@gmail.com>
4 years ago
case bytes.HasPrefix(key, BloomBitsIndexPrefix):
bloomBits.Add(size)
case bytes.HasPrefix(key, skeletonHeaderPrefix) && len(key) == (len(skeletonHeaderPrefix)+8):
beaconHeaders.Add(size)
case bytes.HasPrefix(key, CliqueSnapshotPrefix) && len(key) == 7+common.HashLength:
cliqueSnaps.Add(size)
case bytes.HasPrefix(key, ChtTablePrefix) ||
bytes.HasPrefix(key, ChtIndexTablePrefix) ||
bytes.HasPrefix(key, ChtPrefix): // Canonical hash trie
chtTrieNodes.Add(size)
case bytes.HasPrefix(key, BloomTrieTablePrefix) ||
bytes.HasPrefix(key, BloomTrieIndexPrefix) ||
bytes.HasPrefix(key, BloomTriePrefix): // Bloomtrie sub
bloomTrieNodes.Add(size)
// Verkle trie data is detected, determine the sub-category
case bytes.HasPrefix(key, VerklePrefix):
remain := key[len(VerklePrefix):]
switch {
case IsAccountTrieNode(remain):
verkleTries.Add(size)
case bytes.HasPrefix(remain, stateIDPrefix) && len(remain) == len(stateIDPrefix)+common.HashLength:
verkleStateLookups.Add(size)
case bytes.Equal(remain, persistentStateIDKey):
metadata.Add(size)
case bytes.Equal(remain, trieJournalKey):
metadata.Add(size)
case bytes.Equal(remain, snapSyncStatusFlagKey):
metadata.Add(size)
default:
unaccounted.Add(size)
}
default:
var accounted bool
all: bloom-filter based pruning mechanism (#21724) * cmd, core, tests: initial state pruner core: fix db inspector cmd/geth: add verify-state cmd/geth: add verification tool core/rawdb: implement flatdb cmd, core: fix rebase core/state: use new contract code layout core/state/pruner: avoid deleting genesis state cmd/geth: add helper function core, cmd: fix extract genesis core: minor fixes contracts: remove useless core/state/snapshot: plugin stacktrie core: polish core/state/snapshot: iterate storage concurrently core/state/snapshot: fix iteration core: add comments core/state/snapshot: polish code core/state: polish core/state/snapshot: rebase core/rawdb: add comments core/rawdb: fix tests core/rawdb: improve tests core/state/snapshot: fix concurrent iteration core/state: run pruning during the recovery core, trie: implement martin's idea core, eth: delete flatdb and polish pruner trie: fix import core/state/pruner: add log core/state/pruner: fix issues core/state/pruner: don't read back core/state/pruner: fix contract code write core/state/pruner: check root node presence cmd, core: polish log core/state: use HEAD-127 as the target core/state/snapshot: improve tests cmd/geth: fix verification tool cmd/geth: use HEAD as the verification default target all: replace the bloomfilter with martin's fork cmd, core: polish code core, cmd: forcibly delete state root core/state/pruner: add hash64 core/state/pruner: fix blacklist core/state: remove blacklist cmd, core: delete trie clean cache before pruning cmd, core: fix lint cmd, core: fix rebase core/state: fix the special case for clique networks core/state/snapshot: remove useless code core/state/pruner: capping the snapshot after pruning cmd, core, eth: fixes core/rawdb: update db inspector cmd/geth: polish code core/state/pruner: fsync bloom filter cmd, core: print warning log core/state/pruner: adjust the parameters for bloom filter cmd, core: create the bloom filter by size core: polish core/state/pruner: sanitize invalid bloomfilter size cmd: address comments cmd/geth: address comments cmd/geth: address comment core/state/pruner: address comments core/state/pruner: rename homedir to datadir cmd, core: address comments core/state/pruner: address comment core/state: address comments core, cmd, tests: address comments core: address comments core/state/pruner: release the iterator after each commit core/state/pruner: improve pruner cmd, core: adjust bloom paramters core/state/pruner: fix lint core/state/pruner: fix tests core: fix rebase core/state/pruner: remove atomic rename core/state/pruner: address comments all: run go mod tidy core/state/pruner: avoid false-positive for the middle state roots core/state/pruner: add checks for middle roots cmd/geth: replace crit with error * core/state/pruner: fix lint * core: drop legacy bloom filter * core/state/snapshot: improve pruner * core/state/snapshot: polish concurrent logs to report ETA vs. hashes * core/state/pruner: add progress report for pruning and compaction too * core: fix snapshot test API * core/state: fix some pruning logs * core/state/pruner: support recovering from bloom flush fail Co-authored-by: Péter Szilágyi <peterke@gmail.com>
4 years ago
for _, meta := range [][]byte{
databaseVersionKey, headHeaderKey, headBlockKey, headFastBlockKey, headFinalizedBlockKey,
lastPivotKey, fastTrieProgressKey, snapshotDisabledKey, SnapshotRootKey, snapshotJournalKey,
snapshotGeneratorKey, snapshotRecoveryKey, txIndexTailKey, fastTxLookupLimitKey,
uncleanShutdownKey, badBlockKey, transitionStatusKey, skeletonSyncStatusKey,
persistentStateIDKey, trieJournalKey, snapshotSyncStatusKey, snapSyncStatusFlagKey,
all: bloom-filter based pruning mechanism (#21724) * cmd, core, tests: initial state pruner core: fix db inspector cmd/geth: add verify-state cmd/geth: add verification tool core/rawdb: implement flatdb cmd, core: fix rebase core/state: use new contract code layout core/state/pruner: avoid deleting genesis state cmd/geth: add helper function core, cmd: fix extract genesis core: minor fixes contracts: remove useless core/state/snapshot: plugin stacktrie core: polish core/state/snapshot: iterate storage concurrently core/state/snapshot: fix iteration core: add comments core/state/snapshot: polish code core/state: polish core/state/snapshot: rebase core/rawdb: add comments core/rawdb: fix tests core/rawdb: improve tests core/state/snapshot: fix concurrent iteration core/state: run pruning during the recovery core, trie: implement martin's idea core, eth: delete flatdb and polish pruner trie: fix import core/state/pruner: add log core/state/pruner: fix issues core/state/pruner: don't read back core/state/pruner: fix contract code write core/state/pruner: check root node presence cmd, core: polish log core/state: use HEAD-127 as the target core/state/snapshot: improve tests cmd/geth: fix verification tool cmd/geth: use HEAD as the verification default target all: replace the bloomfilter with martin's fork cmd, core: polish code core, cmd: forcibly delete state root core/state/pruner: add hash64 core/state/pruner: fix blacklist core/state: remove blacklist cmd, core: delete trie clean cache before pruning cmd, core: fix lint cmd, core: fix rebase core/state: fix the special case for clique networks core/state/snapshot: remove useless code core/state/pruner: capping the snapshot after pruning cmd, core, eth: fixes core/rawdb: update db inspector cmd/geth: polish code core/state/pruner: fsync bloom filter cmd, core: print warning log core/state/pruner: adjust the parameters for bloom filter cmd, core: create the bloom filter by size core: polish core/state/pruner: sanitize invalid bloomfilter size cmd: address comments cmd/geth: address comments cmd/geth: address comment core/state/pruner: address comments core/state/pruner: rename homedir to datadir cmd, core: address comments core/state/pruner: address comment core/state: address comments core, cmd, tests: address comments core: address comments core/state/pruner: release the iterator after each commit core/state/pruner: improve pruner cmd, core: adjust bloom paramters core/state/pruner: fix lint core/state/pruner: fix tests core: fix rebase core/state/pruner: remove atomic rename core/state/pruner: address comments all: run go mod tidy core/state/pruner: avoid false-positive for the middle state roots core/state/pruner: add checks for middle roots cmd/geth: replace crit with error * core/state/pruner: fix lint * core: drop legacy bloom filter * core/state/snapshot: improve pruner * core/state/snapshot: polish concurrent logs to report ETA vs. hashes * core/state/pruner: add progress report for pruning and compaction too * core: fix snapshot test API * core/state: fix some pruning logs * core/state/pruner: support recovering from bloom flush fail Co-authored-by: Péter Szilágyi <peterke@gmail.com>
4 years ago
} {
if bytes.Equal(key, meta) {
metadata.Add(size)
accounted = true
break
}
}
if !accounted {
unaccounted.Add(size)
}
}
count++
if count%1000 == 0 && time.Since(logged) > 8*time.Second {
log.Info("Inspecting database", "count", count, "elapsed", common.PrettyDuration(time.Since(start)))
logged = time.Now()
}
}
// Display the database statistic of key-value store.
stats := [][]string{
{"Key-Value store", "Headers", headers.Size(), headers.Count()},
{"Key-Value store", "Bodies", bodies.Size(), bodies.Count()},
{"Key-Value store", "Receipt lists", receipts.Size(), receipts.Count()},
{"Key-Value store", "Difficulties", tds.Size(), tds.Count()},
{"Key-Value store", "Block number->hash", numHashPairings.Size(), numHashPairings.Count()},
{"Key-Value store", "Block hash->number", hashNumPairings.Size(), hashNumPairings.Count()},
{"Key-Value store", "Transaction index", txLookups.Size(), txLookups.Count()},
{"Key-Value store", "Bloombit index", bloomBits.Size(), bloomBits.Count()},
{"Key-Value store", "Contract codes", codes.Size(), codes.Count()},
{"Key-Value store", "Hash trie nodes", legacyTries.Size(), legacyTries.Count()},
{"Key-Value store", "Path trie state lookups", stateLookups.Size(), stateLookups.Count()},
{"Key-Value store", "Path trie account nodes", accountTries.Size(), accountTries.Count()},
{"Key-Value store", "Path trie storage nodes", storageTries.Size(), storageTries.Count()},
{"Key-Value store", "Verkle trie nodes", verkleTries.Size(), verkleTries.Count()},
{"Key-Value store", "Verkle trie state lookups", verkleStateLookups.Size(), verkleStateLookups.Count()},
{"Key-Value store", "Trie preimages", preimages.Size(), preimages.Count()},
{"Key-Value store", "Account snapshot", accountSnaps.Size(), accountSnaps.Count()},
{"Key-Value store", "Storage snapshot", storageSnaps.Size(), storageSnaps.Count()},
{"Key-Value store", "Beacon sync headers", beaconHeaders.Size(), beaconHeaders.Count()},
{"Key-Value store", "Clique snapshots", cliqueSnaps.Size(), cliqueSnaps.Count()},
{"Key-Value store", "Singleton metadata", metadata.Size(), metadata.Count()},
{"Light client", "CHT trie nodes", chtTrieNodes.Size(), chtTrieNodes.Count()},
{"Light client", "Bloom trie nodes", bloomTrieNodes.Size(), bloomTrieNodes.Count()},
}
// Inspect all registered append-only file store then.
ancients, err := inspectFreezers(db)
if err != nil {
return err
}
for _, ancient := range ancients {
for _, table := range ancient.sizes {
stats = append(stats, []string{
fmt.Sprintf("Ancient store (%s)", strings.Title(ancient.name)),
strings.Title(table.name),
table.size.String(),
fmt.Sprintf("%d", ancient.count()),
})
}
total += ancient.size()
}
table := tablewriter.NewWriter(os.Stdout)
table.SetHeader([]string{"Database", "Category", "Size", "Items"})
table.SetFooter([]string{"", "Total", total.String(), " "})
table.AppendBulk(stats)
table.Render()
if unaccounted.size > 0 {
log.Error("Database contains unaccounted data", "size", unaccounted.size, "count", unaccounted.count)
}
return nil
}
// printChainMetadata prints out chain metadata to stderr.
func printChainMetadata(db ethdb.KeyValueStore) {
fmt.Fprintf(os.Stderr, "Chain metadata\n")
for _, v := range ReadChainMetadata(db) {
fmt.Fprintf(os.Stderr, " %s\n", strings.Join(v, ": "))
}
fmt.Fprintf(os.Stderr, "\n\n")
}
// ReadChainMetadata returns a set of key/value pairs that contains information
// about the database chain status. This can be used for diagnostic purposes
// when investigating the state of the node.
func ReadChainMetadata(db ethdb.KeyValueStore) [][]string {
pp := func(val *uint64) string {
if val == nil {
return "<nil>"
}
return fmt.Sprintf("%d (%#x)", *val, *val)
}
data := [][]string{
{"databaseVersion", pp(ReadDatabaseVersion(db))},
{"headBlockHash", fmt.Sprintf("%v", ReadHeadBlockHash(db))},
{"headFastBlockHash", fmt.Sprintf("%v", ReadHeadFastBlockHash(db))},
{"headHeaderHash", fmt.Sprintf("%v", ReadHeadHeaderHash(db))},
{"lastPivotNumber", pp(ReadLastPivotNumber(db))},
{"len(snapshotSyncStatus)", fmt.Sprintf("%d bytes", len(ReadSnapshotSyncStatus(db)))},
{"snapshotDisabled", fmt.Sprintf("%v", ReadSnapshotDisabled(db))},
{"snapshotJournal", fmt.Sprintf("%d bytes", len(ReadSnapshotJournal(db)))},
{"snapshotRecoveryNumber", pp(ReadSnapshotRecoveryNumber(db))},
{"snapshotRoot", fmt.Sprintf("%v", ReadSnapshotRoot(db))},
{"txIndexTail", pp(ReadTxIndexTail(db))},
}
if b := ReadSkeletonSyncStatus(db); b != nil {
data = append(data, []string{"SkeletonSyncStatus", string(b)})
}
return data
}