[wip/experiment] db/snapshotsync: add option to rebuild commitment when using snap.download.to.block

cmd/utils/flags.go

@@ -664,6 +664,11 @@ var (
 		Usage:   "Download snapshots up to the given block number (exclusive). Disabled by default. Useful for testing and shadow forks.",
 		Aliases: []string{"shadow.fork.block"},
 	}
+	SnapDownloadToBlockWithRebuildCommitmentFlag = cli.BoolFlag{
+		Name:  "snap.download.to.block.with.rebuild.commitment",
+		Usage: "Rebuild commitment domain files based on storage/acc/code. Only used with snap.download.to.block. Needed when all downloaded commitment files are purified.",
+		Value: false,
+	}
 	TorrentVerbosityFlag = cli.IntFlag{
 		Name:  "torrent.verbosity",
 		Value: 1,

db/snapshotsync/snapshotsync.go

@@ -478,7 +478,7 @@ func SyncSnapshots(
 				continue
 			}
 
-			if filterToBlock(p.Name, toBlock, toStep, headerchain) {
+			if filterToBlock(p.Name, toBlock, toStep, headerchain, syncCfg.SnapshotDownloadToBlockWithRebuildCommitment) {
 				continue
 			}
 
@@ -512,7 +512,7 @@ func SyncSnapshots(
 	return nil
 }
 
-func filterToBlock(name string, toBlock uint64, toStep uint64, headerchain bool) bool {
+func filterToBlock(name string, toBlock uint64, toStep uint64, headerchain bool, rebuildCommitment bool) bool {
 	if toBlock == 0 {
 		return false // toBlock filtering is not enabled
 	}
@@ -526,6 +526,9 @@ func filterToBlock(name string, toBlock uint64, toStep uint64, headerchain bool)
 	if strings.HasPrefix(name, "caplin/") {
 		return false // not applicable, caplin files are slot-based
 	}
+	if rebuildCommitment && strings.Contains(name, kv.CommitmentDomain.String()) && strings.HasPrefix(name, "domain") {
+		return true
+	}
 	if stateFile {
 		return fileInfo.To > toStep
 	}

db/state/squeeze.go

@@ -6,7 +6,6 @@ import (
 	"encoding/hex"
 	"errors"
 	"fmt"
-	"github.com/erigontech/erigon/db/version"
 	"math"
 	"os"
 	"path/filepath"
@@ -28,6 +27,7 @@ import (
 	downloadertype "github.com/erigontech/erigon/db/snaptype"
 	"github.com/erigontech/erigon/db/state/execctx"
 	"github.com/erigontech/erigon/db/state/statecfg"
+	"github.com/erigontech/erigon/db/version"
 	"github.com/erigontech/erigon/execution/commitment/commitmentdb"
 )
 
@@ -336,8 +336,8 @@ func CheckCommitmentForPrint(ctx context.Context, rwDb kv.TemporalRwDB) (string,
 // RebuildCommitmentFiles recreates commitment files from existing accounts and storage kv files
 // If some commitment exists, they will be accepted as correct and next kv range will be processed.
 // DB expected to be empty, committed into db keys will be not processed.
-func RebuildCommitmentFiles(ctx context.Context, rwDb kv.TemporalRwDB, txNumsReader *rawdbv3.TxNumsReader, logger log.Logger, squeeze bool) (latestRoot []byte, err error) {
-	a := rwDb.(HasAgg).Agg().(*Aggregator)
+func RebuildCommitmentFiles(ctx context.Context, db kv.TemporalRoDB, txNumsReader *rawdbv3.TxNumsReader, logger log.Logger, squeeze bool) (latestRoot []byte, err error) {
+	a := db.(HasAgg).Agg().(*Aggregator)
 
 	// disable hard alignment; allowing commitment and storage/account to have
 	// different visibleFiles
@@ -468,22 +468,22 @@ func RebuildCommitmentFiles(ctx context.Context, rwDb kv.TemporalRwDB, txNumsRea
 				return true, k
 			}
 
-			rwTx, err := rwDb.BeginTemporalRw(ctx)
+			tx, err := db.BeginTemporalRo(ctx)
 			if err != nil {
 				return nil, err
 			}
-			defer rwTx.Rollback()
+			defer tx.Rollback()
 
-			domains, err := execctx.NewSharedDomains(ctx, rwTx, log.New())
+			domains, err := execctx.NewSharedDomains(ctx, tx, log.New())
 			if err != nil {
 				return nil, err
 			}
 
 			domains.SetBlockNum(blockNum)
 			domains.SetTxNum(lastTxnumInShard - 1)
-			domains.GetCommitmentCtx().SetLimitedHistoryStateReader(rwTx, lastTxnumInShard) // this helps to read state from correct file during commitment
+			domains.GetCommitmentCtx().SetLimitedHistoryStateReader(tx, lastTxnumInShard) // this helps to read state from correct file during commitment
 
-			rebuiltCommit, err = rebuildCommitmentShard(ctx, domains, rwTx, nextKey, &rebuiltCommitment{
+			rebuiltCommit, err = rebuildCommitmentShard(ctx, domains, tx, nextKey, &rebuiltCommitment{
 				StepFrom: shardFrom,
 				StepTo:   shardTo,
 				TxnFrom:  rangeFromTxNum,
@@ -503,7 +503,7 @@ func RebuildCommitmentFiles(ctx context.Context, rwDb kv.TemporalRwDB, txNumsRea
 			a.dirtyFilesLock.Lock()
 			a.recalcVisibleFiles(a.dirtyFilesEndTxNumMinimax())
 			a.dirtyFilesLock.Unlock()
-			rwTx.Rollback()
+			tx.Rollback()
 
 			if shardTo+shardStepsSize > lastShard && shardStepsSize > 1 {
 				shardStepsSize /= 2

execution/stagedsync/stage_snapshots.go

@@ -275,6 +275,17 @@ func DownloadAndIndexSnapshotsIfNeed(s *StageState, ctx context.Context, tx kv.R
 		temporal.ForceReopenAggCtx() // otherwise next stages will not see just-indexed-files
 	}
 
+	if cfg.syncConfig.SnapshotDownloadToBlockWithRebuildCommitment && s.BlockNumber == 0 {
+		txNumsReader := cfg.blockReader.TxnumReader(ctx)
+		_, err = state.RebuildCommitmentFiles(ctx, cfg.db, &txNumsReader, logger, true /*squeeze*/)
+		if err != nil {
+			return fmt.Errorf("failed to rebuild commitment files: %w", err)
+		}
+		if temporal, ok := tx.(*temporal.RwTx); ok {
+			temporal.ForceReopenAggCtx() // otherwise next stages will not see new files
+		}
+	}
+
 	// It's ok to notify before tx.Commit(), because RPCDaemon does read list of files by gRPC (not by reading from db)
 	if cfg.notifier.Events != nil {
 		cfg.notifier.Events.OnNewSnapshot()

node/cli/default_flags.go

@@ -112,6 +112,7 @@ var DefaultFlags = []cli.Flag{
 	&utils.SnapStateStopFlag,
 	&utils.SnapSkipStateSnapshotDownloadFlag,
 	&utils.SnapDownloadToBlockFlag,
+	&utils.SnapDownloadToBlockWithRebuildCommitmentFlag,
 	&utils.DbPageSizeFlag,
 	&utils.DbSizeLimitFlag,
 	&utils.DbWriteMapFlag,

node/cli/flags.go

@@ -294,6 +294,7 @@ func ApplyFlagsForEthConfig(ctx *cli.Context, cfg *ethconfig.Config, logger log.
 
 	if ctx.IsSet(utils.SnapDownloadToBlockFlag.Name) {
 		cfg.Sync.SnapshotDownloadToBlock = ctx.Uint64(utils.SnapDownloadToBlockFlag.Name)
+		cfg.Sync.SnapshotDownloadToBlockWithRebuildCommitment = ctx.Bool(utils.SnapDownloadToBlockWithRebuildCommitmentFlag.Name)
 	}
 
 	if stage := ctx.String(SyncLoopBreakAfterFlag.Name); len(stage) > 0 {

node/ethconfig/config.go

@@ -301,5 +301,7 @@ type Sync struct {
 	MaxReorgDepth            uint64
 	KeepExecutionProofs      bool
 	PersistReceiptsCacheV2   bool
-	SnapshotDownloadToBlock  uint64 // exclusive [0,toBlock)
+
+	SnapshotDownloadToBlock                      uint64 // exclusive [0,toBlock)
+	SnapshotDownloadToBlockWithRebuildCommitment bool
 }