2019-08-06 13:40:28 +03:00
|
|
|
// Copyright 2019 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 snapshot
|
|
|
|
|
|
|
|
import (
|
|
|
|
"bytes"
|
2019-11-26 10:48:29 +03:00
|
|
|
"encoding/binary"
|
2020-11-09 17:03:58 +03:00
|
|
|
"fmt"
|
2019-08-06 13:40:28 +03:00
|
|
|
"math/big"
|
|
|
|
"time"
|
|
|
|
|
2019-11-25 17:30:29 +03:00
|
|
|
"github.com/VictoriaMetrics/fastcache"
|
2019-08-06 13:40:28 +03:00
|
|
|
"github.com/ethereum/go-ethereum/common"
|
2019-11-26 10:48:29 +03:00
|
|
|
"github.com/ethereum/go-ethereum/common/math"
|
2019-08-06 13:40:28 +03:00
|
|
|
"github.com/ethereum/go-ethereum/core/rawdb"
|
|
|
|
"github.com/ethereum/go-ethereum/crypto"
|
|
|
|
"github.com/ethereum/go-ethereum/ethdb"
|
|
|
|
"github.com/ethereum/go-ethereum/log"
|
|
|
|
"github.com/ethereum/go-ethereum/rlp"
|
|
|
|
"github.com/ethereum/go-ethereum/trie"
|
|
|
|
)
|
|
|
|
|
|
|
|
var (
|
|
|
|
// emptyRoot is the known root hash of an empty trie.
|
|
|
|
emptyRoot = common.HexToHash("56e81f171bcc55a6ff8345e692c0f86e5b48e01b996cadc001622fb5e363b421")
|
|
|
|
|
|
|
|
// emptyCode is the known hash of the empty EVM bytecode.
|
|
|
|
emptyCode = crypto.Keccak256Hash(nil)
|
|
|
|
)
|
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// generatorStats is a collection of statistics gathered by the snapshot generator
|
2020-04-29 12:53:08 +03:00
|
|
|
// for logging purposes.
|
2019-11-26 10:48:29 +03:00
|
|
|
type generatorStats struct {
|
|
|
|
wiping chan struct{} // Notification channel if wiping is in progress
|
|
|
|
origin uint64 // Origin prefix where generation started
|
|
|
|
start time.Time // Timestamp when generation started
|
|
|
|
accounts uint64 // Number of accounts indexed
|
|
|
|
slots uint64 // Number of storage slots indexed
|
|
|
|
storage common.StorageSize // Account and storage slot size
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// Log creates an contextual log with the given message and the context pulled
|
|
|
|
// from the internally maintained statistics.
|
2020-08-24 13:22:36 +03:00
|
|
|
func (gs *generatorStats) Log(msg string, root common.Hash, marker []byte) {
|
2019-11-26 10:48:29 +03:00
|
|
|
var ctx []interface{}
|
2020-08-24 13:22:36 +03:00
|
|
|
if root != (common.Hash{}) {
|
|
|
|
ctx = append(ctx, []interface{}{"root", root}...)
|
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Figure out whether we're after or within an account
|
|
|
|
switch len(marker) {
|
|
|
|
case common.HashLength:
|
|
|
|
ctx = append(ctx, []interface{}{"at", common.BytesToHash(marker)}...)
|
|
|
|
case 2 * common.HashLength:
|
|
|
|
ctx = append(ctx, []interface{}{
|
|
|
|
"in", common.BytesToHash(marker[:common.HashLength]),
|
|
|
|
"at", common.BytesToHash(marker[common.HashLength:]),
|
|
|
|
}...)
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Add the usual measurements
|
|
|
|
ctx = append(ctx, []interface{}{
|
|
|
|
"accounts", gs.accounts,
|
|
|
|
"slots", gs.slots,
|
|
|
|
"storage", gs.storage,
|
|
|
|
"elapsed", common.PrettyDuration(time.Since(gs.start)),
|
|
|
|
}...)
|
|
|
|
// Calculate the estimated indexing time based on current stats
|
|
|
|
if len(marker) > 0 {
|
|
|
|
if done := binary.BigEndian.Uint64(marker[:8]) - gs.origin; done > 0 {
|
|
|
|
left := math.MaxUint64 - binary.BigEndian.Uint64(marker[:8])
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
speed := done/uint64(time.Since(gs.start)/time.Millisecond+1) + 1 // +1s to avoid division by zero
|
|
|
|
ctx = append(ctx, []interface{}{
|
|
|
|
"eta", common.PrettyDuration(time.Duration(left/speed) * time.Millisecond),
|
|
|
|
}...)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
log.Info(msg, ctx...)
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// generateSnapshot regenerates a brand new snapshot based on an existing state
|
|
|
|
// database and head block asynchronously. The snapshot is returned immediately
|
|
|
|
// and generation is continued in the background until done.
|
|
|
|
func generateSnapshot(diskdb ethdb.KeyValueStore, triedb *trie.Database, cache int, root common.Hash, wiper chan struct{}) *diskLayer {
|
|
|
|
// Wipe any previously existing snapshot from the database if no wiper is
|
2020-02-24 14:26:34 +03:00
|
|
|
// currently in progress.
|
2019-11-26 10:48:29 +03:00
|
|
|
if wiper == nil {
|
|
|
|
wiper = wipeSnapshot(diskdb, true)
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Create a new disk layer with an initialized state marker at zero
|
|
|
|
rawdb.WriteSnapshotRoot(diskdb, root)
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
base := &diskLayer{
|
2020-03-03 10:10:23 +03:00
|
|
|
diskdb: diskdb,
|
|
|
|
triedb: triedb,
|
|
|
|
root: root,
|
|
|
|
cache: fastcache.New(cache * 1024 * 1024),
|
|
|
|
genMarker: []byte{}, // Initialized but empty!
|
|
|
|
genPending: make(chan struct{}),
|
|
|
|
genAbort: make(chan chan *generatorStats),
|
2019-11-26 10:48:29 +03:00
|
|
|
}
|
|
|
|
go base.generate(&generatorStats{wiping: wiper, start: time.Now()})
|
2020-10-29 22:01:58 +03:00
|
|
|
log.Debug("Start snapshot generation", "root", root)
|
2019-11-26 10:48:29 +03:00
|
|
|
return base
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
|
|
|
|
2020-11-09 17:03:58 +03:00
|
|
|
// journalProgress persists the generator stats into the database to resume later.
|
|
|
|
func journalProgress(db ethdb.KeyValueWriter, marker []byte, stats *generatorStats) {
|
|
|
|
// Write out the generator marker. Note it's a standalone disk layer generator
|
|
|
|
// which is not mixed with journal. It's ok if the generator is persisted while
|
|
|
|
// journal is not.
|
|
|
|
entry := journalGenerator{
|
|
|
|
Done: marker == nil,
|
|
|
|
Marker: marker,
|
|
|
|
}
|
|
|
|
if stats != nil {
|
|
|
|
entry.Wiping = (stats.wiping != nil)
|
|
|
|
entry.Accounts = stats.accounts
|
|
|
|
entry.Slots = stats.slots
|
|
|
|
entry.Storage = uint64(stats.storage)
|
|
|
|
}
|
|
|
|
blob, err := rlp.EncodeToBytes(entry)
|
|
|
|
if err != nil {
|
|
|
|
panic(err) // Cannot happen, here to catch dev errors
|
|
|
|
}
|
|
|
|
var logstr string
|
|
|
|
switch len(marker) {
|
|
|
|
case 0:
|
|
|
|
logstr = "done"
|
|
|
|
case common.HashLength:
|
|
|
|
logstr = fmt.Sprintf("%#x", marker)
|
|
|
|
default:
|
|
|
|
logstr = fmt.Sprintf("%#x:%#x", marker[:common.HashLength], marker[common.HashLength:])
|
|
|
|
}
|
|
|
|
log.Debug("Journalled generator progress", "progress", logstr)
|
|
|
|
rawdb.WriteSnapshotGenerator(db, blob)
|
|
|
|
}
|
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// generate is a background thread that iterates over the state and storage tries,
|
|
|
|
// constructing the state snapshot. All the arguments are purely for statistics
|
2021-01-04 11:07:43 +03:00
|
|
|
// gathering and logging, since the method surfs the blocks as they arrive, often
|
2019-11-26 10:48:29 +03:00
|
|
|
// being restarted.
|
|
|
|
func (dl *diskLayer) generate(stats *generatorStats) {
|
|
|
|
// If a database wipe is in operation, wait until it's done
|
|
|
|
if stats.wiping != nil {
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Wiper running, state snapshotting paused", common.Hash{}, dl.genMarker)
|
2019-11-26 10:48:29 +03:00
|
|
|
select {
|
|
|
|
// If wiper is done, resume normal mode of operation
|
|
|
|
case <-stats.wiping:
|
|
|
|
stats.wiping = nil
|
|
|
|
stats.start = time.Now()
|
|
|
|
|
2020-10-05 11:52:36 +03:00
|
|
|
// If generator was aborted during wipe, return
|
2019-11-26 10:48:29 +03:00
|
|
|
case abort := <-dl.genAbort:
|
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Create an account and state iterator pointing to the current generator marker
|
|
|
|
accTrie, err := trie.NewSecure(dl.root, dl.triedb)
|
2019-08-06 13:40:28 +03:00
|
|
|
if err != nil {
|
2019-11-26 10:48:29 +03:00
|
|
|
// The account trie is missing (GC), surf the chain until one becomes available
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Trie missing, state snapshotting paused", dl.root, dl.genMarker)
|
2019-11-26 10:48:29 +03:00
|
|
|
|
|
|
|
abort := <-dl.genAbort
|
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Resuming state snapshot generation", dl.root, dl.genMarker)
|
2019-11-26 10:48:29 +03:00
|
|
|
|
|
|
|
var accMarker []byte
|
|
|
|
if len(dl.genMarker) > 0 { // []byte{} is the start, use nil for that
|
|
|
|
accMarker = dl.genMarker[:common.HashLength]
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
accIt := trie.NewIterator(accTrie.NodeIterator(accMarker))
|
|
|
|
batch := dl.diskdb.NewBatch()
|
|
|
|
|
|
|
|
// Iterate from the previous marker and continue generating the state snapshot
|
|
|
|
logged := time.Now()
|
2019-08-06 13:40:28 +03:00
|
|
|
for accIt.Next() {
|
2019-11-26 10:48:29 +03:00
|
|
|
// Retrieve the current account and flatten it into the internal format
|
|
|
|
accountHash := common.BytesToHash(accIt.Key)
|
|
|
|
|
2019-08-06 13:40:28 +03:00
|
|
|
var acc struct {
|
|
|
|
Nonce uint64
|
|
|
|
Balance *big.Int
|
|
|
|
Root common.Hash
|
|
|
|
CodeHash []byte
|
|
|
|
}
|
|
|
|
if err := rlp.DecodeBytes(accIt.Value, &acc); err != nil {
|
2019-11-26 10:48:29 +03:00
|
|
|
log.Crit("Invalid account encountered during snapshot creation", "err", err)
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2020-04-29 12:53:08 +03:00
|
|
|
data := SlimAccountRLP(acc.Nonce, acc.Balance, acc.Root, acc.CodeHash)
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// If the account is not yet in-progress, write it out
|
|
|
|
if accMarker == nil || !bytes.Equal(accountHash[:], accMarker) {
|
|
|
|
rawdb.WriteAccountSnapshot(batch, accountHash, data)
|
|
|
|
stats.storage += common.StorageSize(1 + common.HashLength + len(data))
|
|
|
|
stats.accounts++
|
|
|
|
}
|
|
|
|
// If we've exceeded our batch allowance or termination was requested, flush to disk
|
|
|
|
var abort chan *generatorStats
|
|
|
|
select {
|
|
|
|
case abort = <-dl.genAbort:
|
|
|
|
default:
|
|
|
|
}
|
|
|
|
if batch.ValueSize() > ethdb.IdealBatchSize || abort != nil {
|
|
|
|
// Only write and set the marker if we actually did something useful
|
|
|
|
if batch.ValueSize() > 0 {
|
2020-11-09 17:03:58 +03:00
|
|
|
// Ensure the generator entry is in sync with the data
|
|
|
|
marker := accountHash[:]
|
|
|
|
journalProgress(batch, marker, stats)
|
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
batch.Write()
|
|
|
|
batch.Reset()
|
|
|
|
|
|
|
|
dl.lock.Lock()
|
2020-11-09 17:03:58 +03:00
|
|
|
dl.genMarker = marker
|
2019-11-26 10:48:29 +03:00
|
|
|
dl.lock.Unlock()
|
|
|
|
}
|
|
|
|
if abort != nil {
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Aborting state snapshot generation", dl.root, accountHash[:])
|
2019-11-26 10:48:29 +03:00
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// If the account is in-progress, continue where we left off (otherwise iterate all)
|
2019-08-06 13:40:28 +03:00
|
|
|
if acc.Root != emptyRoot {
|
2019-11-26 10:48:29 +03:00
|
|
|
storeTrie, err := trie.NewSecure(acc.Root, dl.triedb)
|
2019-08-06 13:40:28 +03:00
|
|
|
if err != nil {
|
2020-12-14 12:27:15 +03:00
|
|
|
log.Error("Generator failed to access storage trie", "root", dl.root, "account", accountHash, "stroot", acc.Root, "err", err)
|
2020-10-05 11:52:36 +03:00
|
|
|
abort := <-dl.genAbort
|
|
|
|
abort <- stats
|
|
|
|
return
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
var storeMarker []byte
|
|
|
|
if accMarker != nil && bytes.Equal(accountHash[:], accMarker) && len(dl.genMarker) > common.HashLength {
|
|
|
|
storeMarker = dl.genMarker[common.HashLength:]
|
|
|
|
}
|
|
|
|
storeIt := trie.NewIterator(storeTrie.NodeIterator(storeMarker))
|
2019-08-06 13:40:28 +03:00
|
|
|
for storeIt.Next() {
|
2019-10-04 16:24:01 +03:00
|
|
|
rawdb.WriteStorageSnapshot(batch, accountHash, common.BytesToHash(storeIt.Key), storeIt.Value)
|
2019-11-26 10:48:29 +03:00
|
|
|
stats.storage += common.StorageSize(1 + 2*common.HashLength + len(storeIt.Value))
|
|
|
|
stats.slots++
|
|
|
|
|
|
|
|
// If we've exceeded our batch allowance or termination was requested, flush to disk
|
|
|
|
var abort chan *generatorStats
|
|
|
|
select {
|
|
|
|
case abort = <-dl.genAbort:
|
|
|
|
default:
|
|
|
|
}
|
|
|
|
if batch.ValueSize() > ethdb.IdealBatchSize || abort != nil {
|
|
|
|
// Only write and set the marker if we actually did something useful
|
|
|
|
if batch.ValueSize() > 0 {
|
2020-11-09 17:03:58 +03:00
|
|
|
// Ensure the generator entry is in sync with the data
|
|
|
|
marker := append(accountHash[:], storeIt.Key...)
|
|
|
|
journalProgress(batch, marker, stats)
|
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
batch.Write()
|
|
|
|
batch.Reset()
|
|
|
|
|
|
|
|
dl.lock.Lock()
|
2020-11-09 17:03:58 +03:00
|
|
|
dl.genMarker = marker
|
2019-11-26 10:48:29 +03:00
|
|
|
dl.lock.Unlock()
|
|
|
|
}
|
|
|
|
if abort != nil {
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Aborting state snapshot generation", dl.root, append(accountHash[:], storeIt.Key...))
|
2019-11-26 10:48:29 +03:00
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
|
|
|
}
|
2020-10-05 11:52:36 +03:00
|
|
|
if err := storeIt.Err; err != nil {
|
|
|
|
log.Error("Generator failed to iterate storage trie", "accroot", dl.root, "acchash", common.BytesToHash(accIt.Key), "stroot", acc.Root, "err", err)
|
|
|
|
abort := <-dl.genAbort
|
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
|
|
|
if time.Since(logged) > 8*time.Second {
|
2020-08-24 13:22:36 +03:00
|
|
|
stats.Log("Generating state snapshot", dl.root, accIt.Key)
|
2019-08-06 13:40:28 +03:00
|
|
|
logged = time.Now()
|
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Some account processed, unmark the marker
|
|
|
|
accMarker = nil
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|
2020-10-05 11:52:36 +03:00
|
|
|
if err := accIt.Err; err != nil {
|
|
|
|
log.Error("Generator failed to iterate account trie", "root", dl.root, "err", err)
|
|
|
|
abort := <-dl.genAbort
|
|
|
|
abort <- stats
|
|
|
|
return
|
|
|
|
}
|
2019-11-26 10:48:29 +03:00
|
|
|
// Snapshot fully generated, set the marker to nil
|
|
|
|
if batch.ValueSize() > 0 {
|
2020-11-09 17:03:58 +03:00
|
|
|
// Ensure the generator entry is in sync with the data
|
|
|
|
journalProgress(batch, nil, stats)
|
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
batch.Write()
|
|
|
|
}
|
|
|
|
log.Info("Generated state snapshot", "accounts", stats.accounts, "slots", stats.slots,
|
|
|
|
"storage", stats.storage, "elapsed", common.PrettyDuration(time.Since(stats.start)))
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
dl.lock.Lock()
|
|
|
|
dl.genMarker = nil
|
2020-03-03 10:10:23 +03:00
|
|
|
close(dl.genPending)
|
2019-11-26 10:48:29 +03:00
|
|
|
dl.lock.Unlock()
|
2019-08-06 13:40:28 +03:00
|
|
|
|
2019-11-26 10:48:29 +03:00
|
|
|
// Someone will be looking for us, wait it out
|
|
|
|
abort := <-dl.genAbort
|
|
|
|
abort <- nil
|
2019-08-06 13:40:28 +03:00
|
|
|
}
|