cmd, core, eth, graphql, trie: no persisted clean trie cache file (#27525)

The clean trie cache is persisted periodically, therefore Geth can
quickly warmup the cache in next restart.

However it will reduce the robustness of system. The assumption is
held in Geth that if the parent trie node is present, then the entire
sub-trie associated with the parent are all prensent.

Imagine the scenario that Geth rewinds itself to a past block and
restart, but Geth finds the root node of "future state" in clean
cache then regard this state is present in disk, while is not in fact.

Another example is offline pruning tool. Whenever an offline pruning
is performed, the clean cache file has to be removed to aviod hitting
the root node of "deleted states" in clean cache.

All in all, compare with the minor performance gain, system robustness
is something we care more.
This commit is contained in:
rjl493456442
2023-07-04 10:21:06 +03:00
committed by GitHub
parent 6ca3ef9a7b
commit 59f7b289c3
13 changed files with 59 additions and 186 deletions
+3 -51
View File
@@ -18,22 +18,18 @@ package trie
import (
"errors"
"runtime"
"time"
"github.com/VictoriaMetrics/fastcache"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/ethdb"
"github.com/ethereum/go-ethereum/log"
"github.com/ethereum/go-ethereum/trie/triedb/hashdb"
"github.com/ethereum/go-ethereum/trie/trienode"
)
// Config defines all necessary options for database.
type Config struct {
Cache int // Memory allowance (MB) to use for caching trie nodes in memory
Journal string // Journal of clean cache to survive node restarts
Preimages bool // Flag whether the preimage of trie key is recorded
Cache int // Memory allowance (MB) to use for caching trie nodes in memory
Preimages bool // Flag whether the preimage of trie key is recorded
}
// backend defines the methods needed to access/update trie nodes in different
@@ -79,11 +75,7 @@ type Database struct {
func prepare(diskdb ethdb.Database, config *Config) *Database {
var cleans *fastcache.Cache
if config != nil && config.Cache > 0 {
if config.Journal == "" {
cleans = fastcache.New(config.Cache * 1024 * 1024)
} else {
cleans = fastcache.LoadFromFileOrNew(config.Journal, config.Cache*1024*1024)
}
cleans = fastcache.New(config.Cache * 1024 * 1024)
}
var preimages *preimageStore
if config != nil && config.Preimages {
@@ -179,46 +171,6 @@ func (db *Database) WritePreimages() {
}
}
// saveCache saves clean state cache to given directory path
// using specified CPU cores.
func (db *Database) saveCache(dir string, threads int) error {
if db.cleans == nil {
return nil
}
log.Info("Writing clean trie cache to disk", "path", dir, "threads", threads)
start := time.Now()
err := db.cleans.SaveToFileConcurrent(dir, threads)
if err != nil {
log.Error("Failed to persist clean trie cache", "error", err)
return err
}
log.Info("Persisted the clean trie cache", "path", dir, "elapsed", common.PrettyDuration(time.Since(start)))
return nil
}
// SaveCache atomically saves fast cache data to the given dir using all
// available CPU cores.
func (db *Database) SaveCache(dir string) error {
return db.saveCache(dir, runtime.GOMAXPROCS(0))
}
// SaveCachePeriodically atomically saves fast cache data to the given dir with
// the specified interval. All dump operation will only use a single CPU core.
func (db *Database) SaveCachePeriodically(dir string, interval time.Duration, stopCh <-chan struct{}) {
ticker := time.NewTicker(interval)
defer ticker.Stop()
for {
select {
case <-ticker.C:
db.saveCache(dir, 1)
case <-stopCh:
return
}
}
}
// Cap iteratively flushes old but still referenced trie nodes until the total
// memory usage goes below the given threshold. The held pre-images accumulated
// up to this point will be flushed in case the size exceeds the threshold.