go-ethereum/triedb/pathdb/disklayer.go
rjl493456442 21920207e4
Some checks are pending
/ Linux Build (push) Waiting to run
/ Linux Build (arm) (push) Waiting to run
/ Docker Image (push) Waiting to run
triedb/pathdb, eth: use double-buffer mechanism in pathdb (#30464)
Previously, PathDB used a single buffer to aggregate database writes,
which needed to be flushed atomically. However, flushing large amounts
of data (e.g., 256MB) caused significant overhead, often blocking the
system for around 3 seconds during the flush.

To mitigate this overhead and reduce performance spikes, a double-buffer
mechanism is introduced. When the active buffer fills up, it is marked
as frozen and a background flushing process is triggered. Meanwhile, a
new buffer is allocated for incoming writes, allowing operations to
continue uninterrupted.

This approach reduces system blocking times and provides flexibility in
adjusting buffer parameters for improved performance.
2025-06-22 20:40:54 +08:00

611 lines
20 KiB
Go

// Copyright 2022 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 pathdb
import (
"bytes"
"fmt"
"sync"
"time"
"github.com/VictoriaMetrics/fastcache"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/core/rawdb"
"github.com/ethereum/go-ethereum/crypto"
"github.com/ethereum/go-ethereum/log"
)
// diskLayer is a low level persistent layer built on top of a key-value store.
type diskLayer struct {
root common.Hash // Immutable, root hash to which this layer was made for
id uint64 // Immutable, corresponding state id
db *Database // Path-based trie database
// These two caches must be maintained separately, because the key
// for the root node of the storage trie (accountHash) is identical
// to the key for the account data.
nodes *fastcache.Cache // GC friendly memory cache of clean nodes
states *fastcache.Cache // GC friendly memory cache of clean states
buffer *buffer // Live buffer to aggregate writes
frozen *buffer // Frozen node buffer waiting for flushing
stale bool // Signals that the layer became stale (state progressed)
lock sync.RWMutex // Lock used to protect stale flag and genMarker
// The generator is set if the state snapshot was not fully completed,
// regardless of whether the background generation is running or not.
// It should only be unset if the generation completes.
generator *generator
}
// newDiskLayer creates a new disk layer based on the passing arguments.
func newDiskLayer(root common.Hash, id uint64, db *Database, nodes *fastcache.Cache, states *fastcache.Cache, buffer *buffer, frozen *buffer) *diskLayer {
// Initialize the clean caches if the memory allowance is not zero
// or reuse the provided caches if they are not nil (inherited from
// the original disk layer).
if nodes == nil && db.config.TrieCleanSize != 0 {
nodes = fastcache.New(db.config.TrieCleanSize)
}
if states == nil && db.config.StateCleanSize != 0 {
states = fastcache.New(db.config.StateCleanSize)
}
return &diskLayer{
root: root,
id: id,
db: db,
nodes: nodes,
states: states,
buffer: buffer,
frozen: frozen,
}
}
// rootHash implements the layer interface, returning root hash of corresponding state.
func (dl *diskLayer) rootHash() common.Hash {
return dl.root
}
// stateID implements the layer interface, returning the state id of disk layer.
func (dl *diskLayer) stateID() uint64 {
return dl.id
}
// parentLayer implements the layer interface, returning nil as there's no layer
// below the disk.
func (dl *diskLayer) parentLayer() layer {
return nil
}
// setGenerator links the given generator to disk layer, representing the
// associated state snapshot is not fully completed yet and the generation
// is potentially running in the background.
func (dl *diskLayer) setGenerator(generator *generator) {
dl.generator = generator
}
// markStale sets the stale flag as true.
func (dl *diskLayer) markStale() {
dl.lock.Lock()
defer dl.lock.Unlock()
if dl.stale {
panic("triedb disk layer is stale") // we've committed into the same base from two children, boom
}
dl.stale = true
}
// node implements the layer interface, retrieving the trie node with the
// provided node info. No error will be returned if the node is not found.
func (dl *diskLayer) node(owner common.Hash, path []byte, depth int) ([]byte, common.Hash, *nodeLoc, error) {
dl.lock.RLock()
defer dl.lock.RUnlock()
if dl.stale {
return nil, common.Hash{}, nil, errSnapshotStale
}
// Try to retrieve the trie node from the not-yet-written node buffer first
// (both the live one and the frozen one). Note the buffer is lock free since
// it's impossible to mutate the buffer before tagging the layer as stale.
for _, buffer := range []*buffer{dl.buffer, dl.frozen} {
if buffer != nil {
n, found := buffer.node(owner, path)
if found {
dirtyNodeHitMeter.Mark(1)
dirtyNodeReadMeter.Mark(int64(len(n.Blob)))
dirtyNodeHitDepthHist.Update(int64(depth))
return n.Blob, n.Hash, &nodeLoc{loc: locDirtyCache, depth: depth}, nil
}
}
}
dirtyNodeMissMeter.Mark(1)
// Try to retrieve the trie node from the clean memory cache
key := nodeCacheKey(owner, path)
if dl.nodes != nil {
if blob := dl.nodes.Get(nil, key); len(blob) > 0 {
cleanNodeHitMeter.Mark(1)
cleanNodeReadMeter.Mark(int64(len(blob)))
return blob, crypto.Keccak256Hash(blob), &nodeLoc{loc: locCleanCache, depth: depth}, nil
}
cleanNodeMissMeter.Mark(1)
}
// Try to retrieve the trie node from the disk.
var blob []byte
if owner == (common.Hash{}) {
blob = rawdb.ReadAccountTrieNode(dl.db.diskdb, path)
} else {
blob = rawdb.ReadStorageTrieNode(dl.db.diskdb, owner, path)
}
// Store the resolved data in the clean cache. The background buffer flusher
// may also write to the clean cache concurrently, but two writers cannot
// write the same item with different content. If the item already exists,
// it will be found in the frozen buffer, eliminating the need to check the
// database.
if dl.nodes != nil && len(blob) > 0 {
dl.nodes.Set(key, blob)
cleanNodeWriteMeter.Mark(int64(len(blob)))
}
return blob, crypto.Keccak256Hash(blob), &nodeLoc{loc: locDiskLayer, depth: depth}, nil
}
// account directly retrieves the account RLP associated with a particular
// hash in the slim data format.
//
// Note the returned account is not a copy, please don't modify it.
func (dl *diskLayer) account(hash common.Hash, depth int) ([]byte, error) {
dl.lock.RLock()
defer dl.lock.RUnlock()
if dl.stale {
return nil, errSnapshotStale
}
// Try to retrieve the trie node from the not-yet-written node buffer first
// (both the live one and the frozen one). Note the buffer is lock free since
// it's impossible to mutate the buffer before tagging the layer as stale.
for _, buffer := range []*buffer{dl.buffer, dl.frozen} {
if buffer != nil {
blob, found := buffer.account(hash)
if found {
dirtyStateHitMeter.Mark(1)
dirtyStateReadMeter.Mark(int64(len(blob)))
dirtyStateHitDepthHist.Update(int64(depth))
if len(blob) == 0 {
stateAccountInexMeter.Mark(1)
} else {
stateAccountExistMeter.Mark(1)
}
return blob, nil
}
}
}
dirtyStateMissMeter.Mark(1)
// If the layer is being generated, ensure the requested account has
// already been covered by the generator.
marker := dl.genMarker()
if marker != nil && bytes.Compare(hash.Bytes(), marker) > 0 {
return nil, errNotCoveredYet
}
// Try to retrieve the account from the memory cache
if dl.states != nil {
if blob, found := dl.states.HasGet(nil, hash[:]); found {
cleanStateHitMeter.Mark(1)
cleanStateReadMeter.Mark(int64(len(blob)))
if len(blob) == 0 {
stateAccountInexMeter.Mark(1)
} else {
stateAccountExistMeter.Mark(1)
}
return blob, nil
}
cleanStateMissMeter.Mark(1)
}
// Try to retrieve the account from the disk.
blob := rawdb.ReadAccountSnapshot(dl.db.diskdb, hash)
// Store the resolved data in the clean cache. The background buffer flusher
// may also write to the clean cache concurrently, but two writers cannot
// write the same item with different content. If the item already exists,
// it will be found in the frozen buffer, eliminating the need to check the
// database.
if dl.states != nil {
dl.states.Set(hash[:], blob)
cleanStateWriteMeter.Mark(int64(len(blob)))
}
if len(blob) == 0 {
stateAccountInexMeter.Mark(1)
stateAccountInexDiskMeter.Mark(1)
} else {
stateAccountExistMeter.Mark(1)
stateAccountExistDiskMeter.Mark(1)
}
return blob, nil
}
// storage directly retrieves the storage data associated with a particular hash,
// within a particular account.
//
// Note the returned account is not a copy, please don't modify it.
func (dl *diskLayer) storage(accountHash, storageHash common.Hash, depth int) ([]byte, error) {
// Hold the lock, ensure the parent won't be changed during the
// state accessing.
dl.lock.RLock()
defer dl.lock.RUnlock()
if dl.stale {
return nil, errSnapshotStale
}
// Try to retrieve the trie node from the not-yet-written node buffer first
// (both the live one and the frozen one). Note the buffer is lock free since
// it's impossible to mutate the buffer before tagging the layer as stale.
for _, buffer := range []*buffer{dl.buffer, dl.frozen} {
if buffer != nil {
if blob, found := buffer.storage(accountHash, storageHash); found {
dirtyStateHitMeter.Mark(1)
dirtyStateReadMeter.Mark(int64(len(blob)))
dirtyStateHitDepthHist.Update(int64(depth))
if len(blob) == 0 {
stateStorageInexMeter.Mark(1)
} else {
stateStorageExistMeter.Mark(1)
}
return blob, nil
}
}
}
dirtyStateMissMeter.Mark(1)
// If the layer is being generated, ensure the requested storage slot
// has already been covered by the generator.
key := append(accountHash[:], storageHash[:]...)
marker := dl.genMarker()
if marker != nil && bytes.Compare(key, marker) > 0 {
return nil, errNotCoveredYet
}
// Try to retrieve the storage slot from the memory cache
if dl.states != nil {
if blob, found := dl.states.HasGet(nil, key); found {
cleanStateHitMeter.Mark(1)
cleanStateReadMeter.Mark(int64(len(blob)))
if len(blob) == 0 {
stateStorageInexMeter.Mark(1)
} else {
stateStorageExistMeter.Mark(1)
}
return blob, nil
}
cleanStateMissMeter.Mark(1)
}
// Try to retrieve the account from the disk
blob := rawdb.ReadStorageSnapshot(dl.db.diskdb, accountHash, storageHash)
// Store the resolved data in the clean cache. The background buffer flusher
// may also write to the clean cache concurrently, but two writers cannot
// write the same item with different content. If the item already exists,
// it will be found in the frozen buffer, eliminating the need to check the
// database.
if dl.states != nil {
dl.states.Set(key, blob)
cleanStateWriteMeter.Mark(int64(len(blob)))
}
if len(blob) == 0 {
stateStorageInexMeter.Mark(1)
stateStorageInexDiskMeter.Mark(1)
} else {
stateStorageExistMeter.Mark(1)
stateStorageExistDiskMeter.Mark(1)
}
return blob, nil
}
// update implements the layer interface, returning a new diff layer on top
// with the given state set.
func (dl *diskLayer) update(root common.Hash, id uint64, block uint64, nodes *nodeSet, states *StateSetWithOrigin) *diffLayer {
return newDiffLayer(dl, root, id, block, nodes, states)
}
// commit merges the given bottom-most diff layer into the node buffer
// and returns a newly constructed disk layer. Note the current disk
// layer must be tagged as stale first to prevent re-access.
func (dl *diskLayer) commit(bottom *diffLayer, force bool) (*diskLayer, error) {
dl.lock.Lock()
defer dl.lock.Unlock()
// Construct and store the state history first. If crash happens after storing
// the state history but without flushing the corresponding states(journal),
// the stored state history will be truncated from head in the next restart.
var (
overflow bool
oldest uint64
)
if dl.db.freezer != nil {
// Bail out with an error if writing the state history fails.
// This can happen, for example, if the device is full.
err := writeHistory(dl.db.freezer, bottom)
if err != nil {
return nil, err
}
// Determine if the persisted history object has exceeded the configured
// limitation, set the overflow as true if so.
tail, err := dl.db.freezer.Tail()
if err != nil {
return nil, err
}
limit := dl.db.config.StateHistory
if limit != 0 && bottom.stateID()-tail > limit {
overflow = true
oldest = bottom.stateID() - limit + 1 // track the id of history **after truncation**
}
}
// Mark the diskLayer as stale before applying any mutations on top.
dl.stale = true
// Store the root->id lookup afterwards. All stored lookups are identified
// by the **unique** state root. It's impossible that in the same chain
// blocks are not adjacent but have the same root.
if dl.id == 0 {
rawdb.WriteStateID(dl.db.diskdb, dl.root, 0)
}
rawdb.WriteStateID(dl.db.diskdb, bottom.rootHash(), bottom.stateID())
// In a unique scenario where the ID of the oldest history object (after tail
// truncation) surpasses the persisted state ID, we take the necessary action
// of forcibly committing the cached dirty states to ensure that the persisted
// state ID remains higher.
persistedID := rawdb.ReadPersistentStateID(dl.db.diskdb)
if !force && persistedID < oldest {
force = true
}
// Merge the trie nodes and flat states of the bottom-most diff layer into the
// buffer as the combined layer.
combined := dl.buffer.commit(bottom.nodes, bottom.states.stateSet)
// Terminate the background state snapshot generation before mutating the
// persistent state.
if combined.full() || force {
// Wait until the previous frozen buffer is fully flushed
if dl.frozen != nil {
if err := dl.frozen.waitFlush(); err != nil {
return nil, err
}
}
// Release the frozen buffer and the internally referenced maps will
// be reclaimed by GC.
dl.frozen = nil
// Terminate the background state snapshot generator before flushing
// to prevent data race.
var (
progress []byte
gen = dl.generator
)
if gen != nil {
gen.stop()
progress = gen.progressMarker()
// If the snapshot has been fully generated, unset the generator
if progress == nil {
dl.setGenerator(nil)
} else {
log.Info("Paused snapshot generation")
}
}
// Freeze the live buffer and schedule background flushing
dl.frozen = combined
dl.frozen.flush(bottom.root, dl.db.diskdb, dl.db.freezer, progress, dl.nodes, dl.states, bottom.stateID(), func() {
// Resume the background generation if it's not completed yet.
// The generator is assumed to be available if the progress is
// not nil.
//
// Notably, the generator will be shared and linked by all the
// disk layer instances, regardless of the generation is terminated
// or not.
if progress != nil {
gen.run(bottom.root)
}
})
// Block until the frozen buffer is fully flushed out if the async flushing
// is not allowed, or if the oldest history surpasses the persisted state ID.
if dl.db.config.NoAsyncFlush || persistedID < oldest {
if err := dl.frozen.waitFlush(); err != nil {
return nil, err
}
dl.frozen = nil
}
combined = newBuffer(dl.db.config.WriteBufferSize, nil, nil, 0)
}
// Link the generator if snapshot is not yet completed
ndl := newDiskLayer(bottom.root, bottom.stateID(), dl.db, dl.nodes, dl.states, combined, dl.frozen)
if dl.generator != nil {
ndl.setGenerator(dl.generator)
}
// To remove outdated history objects from the end, we set the 'tail' parameter
// to 'oldest-1' due to the offset between the freezer index and the history ID.
if overflow {
pruned, err := truncateFromTail(ndl.db.diskdb, ndl.db.freezer, oldest-1)
if err != nil {
return nil, err
}
log.Debug("Pruned state history", "items", pruned, "tailid", oldest)
}
return ndl, nil
}
// revert applies the given state history and return a reverted disk layer.
func (dl *diskLayer) revert(h *history) (*diskLayer, error) {
start := time.Now()
if h.meta.root != dl.rootHash() {
return nil, errUnexpectedHistory
}
if dl.id == 0 {
return nil, fmt.Errorf("%w: zero state id", errStateUnrecoverable)
}
// Apply the reverse state changes upon the current state. This must
// be done before holding the lock in order to access state in "this"
// layer.
nodes, err := apply(dl.db, h.meta.parent, h.meta.root, h.meta.version != stateHistoryV0, h.accounts, h.storages)
if err != nil {
return nil, err
}
// Derive the state modification set from the history, keyed by the hash
// of the account address and the storage key.
accounts, storages := h.stateSet()
// Mark the diskLayer as stale before applying any mutations on top.
dl.lock.Lock()
defer dl.lock.Unlock()
dl.stale = true
// State change may be applied to node buffer, or the persistent
// state, depends on if node buffer is empty or not. If the node
// buffer is not empty, it means that the state transition that
// needs to be reverted is not yet flushed and cached in node
// buffer, otherwise, manipulate persistent state directly.
if !dl.buffer.empty() {
err := dl.buffer.revertTo(dl.db.diskdb, nodes, accounts, storages)
if err != nil {
return nil, err
}
ndl := newDiskLayer(h.meta.parent, dl.id-1, dl.db, dl.nodes, dl.states, dl.buffer, dl.frozen)
// Link the generator if it exists
if dl.generator != nil {
ndl.setGenerator(dl.generator)
}
log.Debug("Reverted data in write buffer", "oldroot", h.meta.root, "newroot", h.meta.parent, "elapsed", common.PrettyDuration(time.Since(start)))
return ndl, nil
}
// Block until the frozen buffer is fully flushed
if dl.frozen != nil {
if err := dl.frozen.waitFlush(); err != nil {
return nil, err
}
// Unset the frozen buffer if it exists, otherwise these "reverted"
// states will still be accessible after revert in frozen buffer.
dl.frozen = nil
}
// Terminate the generator before writing any data to the database.
// This must be done after flushing the frozen buffer, as the generator
// may be restarted at the end of the flush process.
var progress []byte
if dl.generator != nil {
dl.generator.stop()
progress = dl.generator.progressMarker()
}
batch := dl.db.diskdb.NewBatch()
writeNodes(batch, nodes, dl.nodes)
// Provide the original values of modified accounts and storages for revert
writeStates(batch, progress, accounts, storages, dl.states)
rawdb.WritePersistentStateID(batch, dl.id-1)
rawdb.WriteSnapshotRoot(batch, h.meta.parent)
if err := batch.Write(); err != nil {
log.Crit("Failed to write states", "err", err)
}
// Link the generator and resume generation if the snapshot is not yet
// fully completed.
ndl := newDiskLayer(h.meta.parent, dl.id-1, dl.db, dl.nodes, dl.states, dl.buffer, dl.frozen)
if dl.generator != nil && !dl.generator.completed() {
ndl.generator = dl.generator
ndl.generator.run(h.meta.parent)
}
log.Debug("Reverted data in persistent state", "oldroot", h.meta.root, "newroot", h.meta.parent, "elapsed", common.PrettyDuration(time.Since(start)))
return ndl, nil
}
// size returns the approximate size of cached nodes in the disk layer.
func (dl *diskLayer) size() common.StorageSize {
dl.lock.RLock()
defer dl.lock.RUnlock()
if dl.stale {
return 0
}
return common.StorageSize(dl.buffer.size())
}
// resetCache releases the memory held by clean cache to prevent memory leak.
func (dl *diskLayer) resetCache() {
dl.lock.RLock()
defer dl.lock.RUnlock()
// Stale disk layer loses the ownership of clean caches.
if dl.stale {
return
}
if dl.nodes != nil {
dl.nodes.Reset()
}
if dl.states != nil {
dl.states.Reset()
}
}
// genMarker returns the current state snapshot generation progress marker. If
// the state snapshot has already been fully generated, nil is returned.
func (dl *diskLayer) genMarker() []byte {
if dl.generator == nil {
return nil
}
return dl.generator.progressMarker()
}
// genComplete returns a flag indicating whether the state snapshot has been
// fully generated.
func (dl *diskLayer) genComplete() bool {
dl.lock.RLock()
defer dl.lock.RUnlock()
return dl.genMarker() == nil
}
// waitFlush blocks until the background buffer flush is completed.
func (dl *diskLayer) waitFlush() error {
dl.lock.RLock()
defer dl.lock.RUnlock()
if dl.frozen == nil {
return nil
}
return dl.frozen.waitFlush()
}
// terminate releases the frozen buffer if it's not nil and terminates the
// background state generator.
func (dl *diskLayer) terminate() error {
dl.lock.Lock()
defer dl.lock.Unlock()
if dl.frozen != nil {
if err := dl.frozen.waitFlush(); err != nil {
return err
}
dl.frozen = nil
}
if dl.generator != nil {
dl.generator.stop()
}
return nil
}