feat: implement incremental snapshot

cmd/geth/chaincmd.go

@@ -363,7 +363,7 @@ func initGenesis(ctx *cli.Context) error {
 		log.Warn("Multi-database is an experimental feature")
 	}
 
-	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, ctx.Bool(utils.CachePreimagesFlag.Name), false, genesis.IsVerkle())
+	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, ctx.Bool(utils.CachePreimagesFlag.Name), false, genesis.IsVerkle(), false)
 	defer triedb.Close()
 
 	_, hash, compatErr, err := core.SetupGenesisBlockWithOverride(chaindb, triedb, genesis, &overrides)
@@ -1091,7 +1091,7 @@ func dump(ctx *cli.Context) error {
 		return err
 	}
 	defer db.Close()
-	triedb := utils.MakeTrieDatabase(ctx, stack, db, true, true, false) // always enable preimage lookup
+	triedb := utils.MakeTrieDatabase(ctx, stack, db, true, true, false, false) // always enable preimage lookup
 	defer triedb.Close()
 
 	state, err := state.New(root, state.NewDatabase(triedb, nil))

cmd/geth/dbcmd.go

@@ -93,6 +93,7 @@ Remove blockchain and state databases`,
 			dbTrieDeleteCmd,
 			dbDeleteTrieStateCmd,
 			ancientInspectCmd,
+			incrInspectCmd,
 		},
 	}
 	dbInspectCmd = &cli.Command{
@@ -275,6 +276,13 @@ of ancientStore, will also displays the reserved number of blocks in ancientStor
 		}, utils.NetworkFlags, utils.DatabaseFlags),
 		Description: "This command queries the history of the account or storage slot within the specified block range",
 	}
+	incrInspectCmd = &cli.Command{
+		Action:      inspectIncrSnapshot,
+		Name:        "inspect-incr-snapshot",
+		Flags:       []cli.Flag{utils.IncrSnapshotPathFlag},
+		Usage:       "Inspect the incremental snapshot information",
+		Description: `This command reads and displays incremental store information`,
+	}
 )
 
 func removeDB(ctx *cli.Context) error {
@@ -945,7 +953,7 @@ func dbDumpTrie(ctx *cli.Context) error {
 
 	db := utils.MakeChainDatabase(ctx, stack, true)
 	defer db.Close()
-	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, true, false, false)
 	defer triedb.Close()
 
 	var (
@@ -1279,7 +1287,7 @@ func inspectHistory(ctx *cli.Context) error {
 	db := utils.MakeChainDatabase(ctx, stack, true)
 	defer db.Close()
 
-	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, false, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, false, false, false)
 	defer triedb.Close()
 
 	var (
@@ -1327,3 +1335,14 @@ func inspectHistory(ctx *cli.Context) error {
 	}
 	return inspectStorage(triedb, start, end, address, slot, ctx.Bool("raw"))
 }
+
+func inspectIncrSnapshot(ctx *cli.Context) error {
+	if !ctx.IsSet(utils.IncrSnapshotPathFlag.Name) {
+		return errors.New("increment snapshot path is not set")
+	}
+	baseDir := ctx.String(utils.IncrSnapshotPathFlag.Name)
+	if err := rawdb.InspectIncrStore(baseDir); err != nil {
+		return err
+	}
+	return nil
+}

cmd/geth/main.go

@@ -183,6 +183,13 @@ var (
 		utils.LogDebugFlag,
 		utils.LogBacktraceAtFlag,
 		utils.BlobExtraReserveFlag,
+		utils.EnableIncrSnapshotFlag,
+		utils.IncrSnapshotPathFlag,
+		utils.IncrSnapshotBlockIntervalFlag,
+		utils.IncrSnapshotStateBufferFlag,
+		utils.IncrSnapshotKeptBlocksFlag,
+		utils.UseRemoteIncrSnapshotFlag,
+		utils.RemoteIncrSnapshotURLFlag,
 		// utils.BeaconApiFlag,
 		// utils.BeaconApiHeaderFlag,
 		// utils.BeaconThresholdFlag,

cmd/geth/snapshot.go

@@ -27,6 +27,7 @@ import (
 
 	"github.com/ethereum/go-ethereum/cmd/utils"
 	"github.com/ethereum/go-ethereum/common"
+	"github.com/ethereum/go-ethereum/core"
 	"github.com/ethereum/go-ethereum/core/rawdb"
 	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/state/pruner"
@@ -163,6 +164,15 @@ The export-preimages command exports hash preimages to a flat file, in exactly
 the expected order for the overlay tree migration.
 `,
 			},
+			{
+				Action:    mergeIncrSnapshot,
+				Name:      "merge-incr-snapshot",
+				Usage:     "Merge the incremental snapshot into local data",
+				ArgsUsage: "",
+				Flags: slices.Concat([]cli.Flag{utils.IncrSnapshotPathFlag},
+					utils.DatabaseFlags),
+				Description: `This command merges multiple incremental snapshots into local data`,
+			},
 		},
 	}
 )
@@ -220,7 +230,7 @@ func verifyState(ctx *cli.Context) error {
 		log.Error("Failed to load head block")
 		return errors.New("no head block")
 	}
-	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false, false)
 	defer triedb.Close()
 
 	var (
@@ -285,7 +295,7 @@ func traverseState(ctx *cli.Context) error {
 	chaindb := utils.MakeChainDatabase(ctx, stack, true)
 	defer chaindb.Close()
 
-	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false, false)
 	defer triedb.Close()
 
 	headBlock := rawdb.ReadHeadBlock(chaindb)
@@ -394,7 +404,7 @@ func traverseRawState(ctx *cli.Context) error {
 	chaindb := utils.MakeChainDatabase(ctx, stack, true)
 	defer chaindb.Close()
 
-	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false, false)
 	defer triedb.Close()
 
 	headBlock := rawdb.ReadHeadBlock(chaindb)
@@ -562,7 +572,7 @@ func dumpState(ctx *cli.Context) error {
 		return err
 	}
 	defer db.Close()
-	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, db, false, true, false, false)
 	defer triedb.Close()
 
 	snapConfig := snapshot.Config{
@@ -645,7 +655,7 @@ func snapshotExportPreimages(ctx *cli.Context) error {
 	chaindb := utils.MakeChainDatabase(ctx, stack, true)
 	defer chaindb.Close()
 
-	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false)
+	triedb := utils.MakeTrieDatabase(ctx, stack, chaindb, false, true, false, false)
 	defer triedb.Close()
 
 	var root common.Hash
@@ -707,3 +717,70 @@ func checkAccount(ctx *cli.Context) error {
 	log.Info("Checked the snapshot journalled storage", "time", common.PrettyDuration(time.Since(start)))
 	return nil
 }
+
+// mergeIncrSnapshot merges the incremental snapshot into local data.
+func mergeIncrSnapshot(ctx *cli.Context) error {
+	stack, _ := makeConfigNode(ctx)
+	defer stack.Close()
+
+	chainDB := utils.MakeChainDatabase(ctx, stack, false)
+	defer chainDB.Close()
+
+	trieDB := utils.MakeTrieDatabase(ctx, stack, chainDB, false, false, false, true)
+	defer trieDB.Close()
+
+	if !ctx.IsSet(utils.IncrSnapshotPathFlag.Name) {
+		return errors.New("incremental snapshot path is not set")
+	}
+	path := ctx.String(utils.IncrSnapshotPathFlag.Name)
+
+	startBlock, err := trieDB.GetStartBlock()
+	if err != nil {
+		log.Error("Failed to get start block", "error", err)
+		return err
+	}
+	dirs, err := rawdb.GetAllIncrDirs(path)
+	if err != nil {
+		log.Error("Failed to get all incremental directories", "err", err)
+		return err
+	}
+	if startBlock < dirs[0].StartBlockNum {
+		return fmt.Errorf("local start block %d is lower than incr first start block %d", startBlock, dirs[0].StartBlockNum)
+	}
+
+	for i := 1; i < len(dirs); i++ {
+		prevFile := dirs[i-1]
+		currFile := dirs[i]
+
+		expectedStartBlock := prevFile.EndBlockNum + 1
+		if currFile.StartBlockNum != expectedStartBlock {
+			return fmt.Errorf("file continuity broken: file %s ends at %d, but file %s starts at %d (expected %d)",
+				prevFile.Name, prevFile.EndBlockNum, currFile.Name, currFile.StartBlockNum, expectedStartBlock)
+		}
+	}
+
+	log.Info("Start merging incremental snapshot", "path", path, "incremental snapshot number", len(dirs))
+	for i, dir := range dirs {
+		if i == len(dirs)-1 {
+			complete, err := rawdb.CheckIncrSnapshotComplete(dir.Path)
+			if err != nil {
+				log.Error("Failed to check last incr snapshot complete", "err", err)
+				return err
+			}
+			if !complete {
+				log.Warn("Skip last incr snapshot due to data is incomplete")
+				continue
+			}
+		}
+
+		if dir.StartBlockNum >= startBlock && dir.EndBlockNum > startBlock {
+			if err = core.MergeIncrSnapshot(chainDB, trieDB, dir.Path); err != nil {
+				log.Error("Failed to merge incremental snapshot", "err", err)
+				return err
+			}
+		} else {
+			log.Info("Skip merge incremental snapshot", "dir", dir.Name)
+		}
+	}
+	return nil
+}

cmd/utils/flags.go

@@ -1274,6 +1274,50 @@ Please note that --` + MetricsHTTPFlag.Name + ` must be set to start the server.
 		Value:    fakebeacon.DefaultPort,
 		Category: flags.APICategory,
 	}
+
+	// incremental snapshot related flags
+	EnableIncrSnapshotFlag = &cli.BoolFlag{
+		Name:     "incr.enable",
+		Usage:    "Enable incremental snapshot generation",
+		Value:    false,
+		Category: flags.StateCategory,
+	}
+	IncrSnapshotPathFlag = &flags.DirectoryFlag{
+		Name:     "incr.datadir",
+		Usage:    "Data directory for storing incremental snapshot data: can be used to store generated or downloaded incremental snapshot",
+		Value:    "",
+		Category: flags.StateCategory,
+	}
+	IncrSnapshotBlockIntervalFlag = &cli.Uint64Flag{
+		Name:     "incr.block-interval",
+		Usage:    "Set how many blocks interval are stored into one incremental snapshot",
+		Value:    pathdb.DefaultBlockInterval,
+		Category: flags.StateCategory,
+	}
+	IncrSnapshotStateBufferFlag = &cli.Uint64Flag{
+		Name:     "incr.state-buffer",
+		Usage:    "Set the incr state memory buffer to aggregate MPT trie nodes. The larger the setting, the smaller the incr snapshot size",
+		Value:    pathdb.DefaultIncrStateBufferSize,
+		Category: flags.StateCategory,
+	}
+	IncrSnapshotKeptBlocksFlag = &cli.Uint64Flag{
+		Name:     "incr.kept-blocks",
+		Usage:    "Set how many blocks are kept in incr snapshot. At least is 1024 blocks",
+		Value:    pathdb.DefaultKeptBlocks,
+		Category: flags.StateCategory,
+	}
+	UseRemoteIncrSnapshotFlag = &cli.BoolFlag{
+		Name:     "incr.use-remote",
+		Usage:    "Enable download and merge incremental snapshots into local data",
+		Value:    false,
+		Category: flags.StateCategory,
+	}
+	RemoteIncrSnapshotURLFlag = &cli.StringFlag{
+		Name:     "incr.remote-url",
+		Usage:    "Set from which remote url is used to download incremental snapshots",
+		Value:    "",
+		Category: flags.StateCategory,
+	}
 )
 
 var (
@@ -2318,6 +2362,38 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *ethconfig.Config) {
 			cfg.VMTraceJsonConfig = ctx.String(VMTraceJsonConfigFlag.Name)
 		}
 	}
+
+	// Download and merge incremental snapshot config
+	if ctx.IsSet(UseRemoteIncrSnapshotFlag.Name) {
+		cfg.UseRemoteIncrSnapshot = true
+		if !ctx.IsSet(RemoteIncrSnapshotURLFlag.Name) {
+			Fatalf("Must provide a remote increment snapshot URL")
+		} else {
+			cfg.RemoteIncrSnapshotURL = ctx.String(RemoteIncrSnapshotURLFlag.Name)
+		}
+		if ctx.IsSet(IncrSnapshotPathFlag.Name) {
+			cfg.IncrSnapshotPath = ctx.String(IncrSnapshotPathFlag.Name)
+		} else {
+			Fatalf("Must provide a path to store downloaded incr snapshot")
+		}
+	}
+
+	// enable incremental snapshot generation config
+	if ctx.IsSet(EnableIncrSnapshotFlag.Name) {
+		cfg.EnableIncrSnapshots = true
+		if ctx.IsSet(IncrSnapshotPathFlag.Name) {
+			cfg.IncrSnapshotPath = ctx.String(IncrSnapshotPathFlag.Name)
+		}
+		if ctx.IsSet(IncrSnapshotBlockIntervalFlag.Name) {
+			cfg.IncrSnapshotBlockInterval = ctx.Uint64(IncrSnapshotBlockIntervalFlag.Name)
+		}
+		if ctx.IsSet(IncrSnapshotStateBufferFlag.Name) {
+			cfg.IncrSnapshotStateBuffer = ctx.Uint64(IncrSnapshotStateBufferFlag.Name)
+		}
+		if ctx.IsSet(IncrSnapshotKeptBlocksFlag.Name) {
+			cfg.IncrSnapshotKeptBlocks = ctx.Uint64(IncrSnapshotKeptBlocksFlag.Name)
+		}
+	}
 }
 
 // SetDNSDiscoveryDefaults configures DNS discovery with the given URL if
@@ -2783,7 +2859,7 @@ func MakeConsolePreloads(ctx *cli.Context) []string {
 }
 
 // MakeTrieDatabase constructs a trie database based on the configured scheme.
-func MakeTrieDatabase(ctx *cli.Context, stack *node.Node, disk ethdb.Database, preimage bool, readOnly bool, isVerkle bool) *triedb.Database {
+func MakeTrieDatabase(ctx *cli.Context, stack *node.Node, disk ethdb.Database, preimage bool, readOnly bool, isVerkle bool, mergeIncr bool) *triedb.Database {
 	config := &triedb.Config{
 		Preimages: preimage,
 		IsVerkle:  isVerkle,
@@ -2803,6 +2879,9 @@ func MakeTrieDatabase(ctx *cli.Context, stack *node.Node, disk ethdb.Database, p
 		config.PathDB = pathdb.ReadOnly
 	} else {
 		config.PathDB = pathdb.Defaults
+		if mergeIncr {
+			config.PathDB.MergeIncr = true
+		}
 	}
 	config.PathDB.JournalFilePath = fmt.Sprintf("%s/%s", stack.ResolvePath("chaindata"), eth.JournalFileName)
 	return triedb.NewDatabase(disk, config)