core, trie: intermediate mempool between trie and database (ethereum#…

…15857)

This commit reduces database I/O by not writing every state trie to disk.

accounts/abi/bind/backends/simulated.go

@@ -68,7 +68,7 @@ func NewSimulatedBackend(alloc core.GenesisAlloc) *SimulatedBackend {
 	database, _ := ethdb.NewMemDatabase()
 	genesis := core.Genesis{Config: params.AllEthashProtocolChanges, Alloc: alloc}
 	genesis.MustCommit(database)
-	blockchain, _ := core.NewBlockChain(database, genesis.Config, ethash.NewFaker(), vm.Config{})
+	blockchain, _ := core.NewBlockChain(database, nil, genesis.Config, ethash.NewFaker(), vm.Config{})
 
 	backend := &SimulatedBackend{
 		database:   database,
@@ -102,8 +102,10 @@ func (b *SimulatedBackend) Rollback() {
 
 func (b *SimulatedBackend) rollback() {
 	blocks, _ := core.GenerateChain(b.config, b.blockchain.CurrentBlock(), ethash.NewFaker(), b.database, 1, func(int, *core.BlockGen) {})
+	statedb, _ := b.blockchain.State()
+
 	b.pendingBlock = blocks[0]
-	b.pendingState, _ = state.New(b.pendingBlock.Root(), state.NewDatabase(b.database))
+	b.pendingState, _ = state.New(b.pendingBlock.Root(), statedb.Database())
 }
 
 // CodeAt returns the code associated with a certain account in the blockchain.
@@ -309,8 +311,10 @@ func (b *SimulatedBackend) SendTransaction(ctx context.Context, tx *types.Transa
 		}
 		block.AddTx(tx)
 	})
+	statedb, _ := b.blockchain.State()
+
 	b.pendingBlock = blocks[0]
-	b.pendingState, _ = state.New(b.pendingBlock.Root(), state.NewDatabase(b.database))
+	b.pendingState, _ = state.New(b.pendingBlock.Root(), statedb.Database())
 	return nil
 }
 
@@ -386,8 +390,10 @@ func (b *SimulatedBackend) AdjustTime(adjustment time.Duration) error {
 		}
 		block.OffsetTime(int64(adjustment.Seconds()))
 	})
+	statedb, _ := b.blockchain.State()
+
 	b.pendingBlock = blocks[0]
-	b.pendingState, _ = state.New(b.pendingBlock.Root(), state.NewDatabase(b.database))
+	b.pendingState, _ = state.New(b.pendingBlock.Root(), statedb.Database())
 
 	return nil
 }

cmd/evm/runner.go

@@ -96,7 +96,9 @@ func runCmd(ctx *cli.Context) error {
 	}
 	if ctx.GlobalString(GenesisFlag.Name) != "" {
 		gen := readGenesis(ctx.GlobalString(GenesisFlag.Name))
-		_, statedb = gen.ToBlock()
+		db, _ := ethdb.NewMemDatabase()
+		genesis := gen.ToBlock(db)
+		statedb, _ = state.New(genesis.Root(), state.NewDatabase(db))
 		chainConfig = gen.Config
 	} else {
 		db, _ := ethdb.NewMemDatabase()

cmd/geth/chaincmd.go

@@ -202,7 +202,7 @@ func importChain(ctx *cli.Context) error {
 
 	if len(ctx.Args()) == 1 {
 		if err := utils.ImportChain(chain, ctx.Args().First()); err != nil {
-			utils.Fatalf("Import error: %v", err)
+			log.Error("Import error", "err", err)
 		}
 	} else {
 		for _, arg := range ctx.Args() {
@@ -211,7 +211,7 @@ func importChain(ctx *cli.Context) error {
 			}
 		}
 	}
-
+	chain.Stop()
 	fmt.Printf("Import done in %v.\n\n", time.Since(start))
 
 	// Output pre-compaction stats mostly to see the import trashing

cmd/geth/main.go

@@ -85,10 +85,13 @@ var (
 		utils.FastSyncFlag,
 		utils.LightModeFlag,
 		utils.SyncModeFlag,
+		utils.GCModeFlag,
 		utils.LightServFlag,
 		utils.LightPeersFlag,
 		utils.LightKDFFlag,
 		utils.CacheFlag,
+		utils.CacheDatabaseFlag,
+		utils.CacheGCFlag,
 		utils.TrieCacheGenFlag,
 		utils.ListenPortFlag,
 		utils.MaxPeersFlag,

cmd/geth/usage.go

@@ -22,10 +22,11 @@ import (
 	"io"
 	"sort"
 
+	"strings"
+
 	"github.com/ethereum/go-ethereum/cmd/utils"
 	"github.com/ethereum/go-ethereum/internal/debug"
 	"gopkg.in/urfave/cli.v1"
-	"strings"
 )
 
 // AppHelpTemplate is the test template for the default, global app help topic.
@@ -74,6 +75,7 @@ var AppHelpFlagGroups = []flagGroup{
 			utils.TestnetFlag,
 			utils.RinkebyFlag,
 			utils.SyncModeFlag,
+			utils.GCModeFlag,
 			utils.EthStatsURLFlag,
 			utils.IdentityFlag,
 			utils.LightServFlag,
@@ -127,6 +129,8 @@ var AppHelpFlagGroups = []flagGroup{
 		Name: "PERFORMANCE TUNING",
 		Flags: []cli.Flag{
 			utils.CacheFlag,
+			utils.CacheDatabaseFlag,
+			utils.CacheGCFlag,
 			utils.TrieCacheGenFlag,
 		},
 	},

cmd/utils/cmd.go

@@ -116,7 +116,6 @@ func ImportChain(chain *core.BlockChain, fn string) error {
 			return err
 		}
 	}
-
 	stream := rlp.NewStream(reader, 0)
 
 	// Run actual the import.
@@ -150,25 +149,34 @@ func ImportChain(chain *core.BlockChain, fn string) error {
 		if checkInterrupt() {
 			return fmt.Errorf("interrupted")
 		}
-		if hasAllBlocks(chain, blocks[:i]) {
+		missing := missingBlocks(chain, blocks[:i])
+		if len(missing) == 0 {
 			log.Info("Skipping batch as all blocks present", "batch", batch, "first", blocks[0].Hash(), "last", blocks[i-1].Hash())
 			continue
 		}
-
-		if _, err := chain.InsertChain(blocks[:i]); err != nil {
+		if _, err := chain.InsertChain(missing); err != nil {
 			return fmt.Errorf("invalid block %d: %v", n, err)
 		}
 	}
 	return nil
 }
 
-func hasAllBlocks(chain *core.BlockChain, bs []*types.Block) bool {
-	for _, b := range bs {
-		if !chain.HasBlock(b.Hash(), b.NumberU64()) {
-			return false
+func missingBlocks(chain *core.BlockChain, blocks []*types.Block) []*types.Block {
+	head := chain.CurrentBlock()
+	for i, block := range blocks {
+		// If we're behind the chain head, only check block, state is available at head
+		if head.NumberU64() > block.NumberU64() {
+			if !chain.HasBlock(block.Hash(), block.NumberU64()) {
+				return blocks[i:]
+			}
+			continue
+		}
+		// If we're above the chain head, state availability is a must
+		if !chain.HasBlockAndState(block.Hash(), block.NumberU64()) {
+			return blocks[i:]
 		}
 	}
-	return true
+	return nil
 }
 
 func ExportChain(blockchain *core.BlockChain, fn string) error {

cmd/utils/flags.go

@@ -170,7 +170,11 @@ var (
 		Usage: `Blockchain sync mode ("fast", "full", or "light")`,
 		Value: &defaultSyncMode,
 	}
-
+	GCModeFlag = cli.StringFlag{
+		Name:  "gcmode",
+		Usage: `Blockchain garbage collection mode ("full", "archive")`,
+		Value: "full",
+	}
 	LightServFlag = cli.IntFlag{
 		Name:  "lightserv",
 		Usage: "Maximum percentage of time allowed for serving LES requests (0-90)",
@@ -293,8 +297,18 @@ var (
 	// Performance tuning settings
 	CacheFlag = cli.IntFlag{
 		Name:  "cache",
-		Usage: "Megabytes of memory allocated to internal caching (min 16MB / database forced)",
-		Value: 128,
+		Usage: "Megabytes of memory allocated to internal caching",
+		Value: 1024,
+	}
+	CacheDatabaseFlag = cli.IntFlag{
+		Name:  "cache.database",
+		Usage: "Percentage of cache memory allowance to use for database io",
+		Value: 75,
+	}
+	CacheGCFlag = cli.IntFlag{
+		Name:  "cache.gc",
+		Usage: "Percentage of cache memory allowance to use for trie pruning",
+		Value: 25,
 	}
 	TrieCacheGenFlag = cli.IntFlag{
 		Name:  "trie-cache-gens",
@@ -1021,11 +1035,19 @@ func SetEthConfig(ctx *cli.Context, stack *node.Node, cfg *eth.Config) {
 		cfg.NetworkId = ctx.GlobalUint64(NetworkIdFlag.Name)
 	}
 
-	if ctx.GlobalIsSet(CacheFlag.Name) {
-		cfg.DatabaseCache = ctx.GlobalInt(CacheFlag.Name)
+	if ctx.GlobalIsSet(CacheFlag.Name) || ctx.GlobalIsSet(CacheDatabaseFlag.Name) {
+		cfg.DatabaseCache = ctx.GlobalInt(CacheFlag.Name) * ctx.GlobalInt(CacheDatabaseFlag.Name) / 100
 	}
 	cfg.DatabaseHandles = makeDatabaseHandles()
 
+	if gcmode := ctx.GlobalString(GCModeFlag.Name); gcmode != "full" && gcmode != "archive" {
+		Fatalf("--%s must be either 'full' or 'archive'", GCModeFlag.Name)
+	}
+	cfg.NoPruning = ctx.GlobalString(GCModeFlag.Name) == "archive"
+
+	if ctx.GlobalIsSet(CacheFlag.Name) || ctx.GlobalIsSet(CacheGCFlag.Name) {
+		cfg.TrieCache = ctx.GlobalInt(CacheFlag.Name) * ctx.GlobalInt(CacheGCFlag.Name) / 100
+	}
 	if ctx.GlobalIsSet(MinerThreadsFlag.Name) {
 		cfg.MinerThreads = ctx.GlobalInt(MinerThreadsFlag.Name)
 	}
@@ -1157,7 +1179,7 @@ func SetupNetwork(ctx *cli.Context) {
 // MakeChainDatabase open an LevelDB using the flags passed to the client and will hard crash if it fails.
 func MakeChainDatabase(ctx *cli.Context, stack *node.Node) ethdb.Database {
 	var (
-		cache   = ctx.GlobalInt(CacheFlag.Name)
+		cache   = ctx.GlobalInt(CacheFlag.Name) * ctx.GlobalInt(CacheDatabaseFlag.Name) / 100
 		handles = makeDatabaseHandles()
 	)
 	name := "chaindata"
@@ -1209,8 +1231,19 @@ func MakeChain(ctx *cli.Context, stack *node.Node) (chain *core.BlockChain, chai
 			})
 		}
 	}
+	if gcmode := ctx.GlobalString(GCModeFlag.Name); gcmode != "full" && gcmode != "archive" {
+		Fatalf("--%s must be either 'full' or 'archive'", GCModeFlag.Name)
+	}
+	cache := &core.CacheConfig{
+		Disabled:      ctx.GlobalString(GCModeFlag.Name) == "archive",
+		TrieNodeLimit: eth.DefaultConfig.TrieCache,
+		TrieTimeLimit: eth.DefaultConfig.TrieTimeout,
+	}
+	if ctx.GlobalIsSet(CacheFlag.Name) || ctx.GlobalIsSet(CacheGCFlag.Name) {
+		cache.TrieNodeLimit = ctx.GlobalInt(CacheFlag.Name) * ctx.GlobalInt(CacheGCFlag.Name) / 100
+	}
 	vmcfg := vm.Config{EnablePreimageRecording: ctx.GlobalBool(VMEnableDebugFlag.Name)}
-	chain, err = core.NewBlockChain(chainDb, config, engine, vmcfg)
+	chain, err = core.NewBlockChain(chainDb, cache, config, engine, vmcfg)
 	if err != nil {
 		Fatalf("Can't create BlockChain: %v", err)
 	}

common/size.go

@@ -20,18 +20,29 @@ import (
 	"fmt"
 )
 
+// StorageSize is a wrapper around a float value that supports user friendly
+// formatting.
 type StorageSize float64
 
-func (self StorageSize) String() string {
-	if self > 1000000 {
-		return fmt.Sprintf("%.2f mB", self/1000000)
-	} else if self > 1000 {
-		return fmt.Sprintf("%.2f kB", self/1000)
+// String implements the stringer interface.
+func (s StorageSize) String() string {
+	if s > 1000000 {
+		return fmt.Sprintf("%.2f mB", s/1000000)
+	} else if s > 1000 {
+		return fmt.Sprintf("%.2f kB", s/1000)
 	} else {
-		return fmt.Sprintf("%.2f B", self)
+		return fmt.Sprintf("%.2f B", s)
 	}
 }
 
-func (self StorageSize) Int64() int64 {
-	return int64(self)
+// TerminalString implements log.TerminalStringer, formatting a string for console
+// output during logging.
+func (s StorageSize) TerminalString() string {
+	if s > 1000000 {
+		return fmt.Sprintf("%.2fmB", s/1000000)
+	} else if s > 1000 {
+		return fmt.Sprintf("%.2fkB", s/1000)
+	} else {
+		return fmt.Sprintf("%.2fB", s)
+	}
 }

consensus/errors.go

@@ -23,6 +23,10 @@ var (
 	// that is unknown.
 	ErrUnknownAncestor = errors.New("unknown ancestor")
 
+	// ErrPrunedAncestor is returned when validating a block requires an ancestor
+	// that is known, but the state of which is not available.
+	ErrPrunedAncestor = errors.New("pruned ancestor")
+
 	// ErrFutureBlock is returned when a block's timestamp is in the future according
 	// to the current node.
 	ErrFutureBlock = errors.New("block in the future")

core/bench_test.go

@@ -173,7 +173,7 @@ func benchInsertChain(b *testing.B, disk bool, gen func(int, *BlockGen)) {
 
 	// Time the insertion of the new chain.
 	// State and blocks are stored in the same DB.
-	chainman, _ := NewBlockChain(db, gspec.Config, ethash.NewFaker(), vm.Config{})
+	chainman, _ := NewBlockChain(db, nil, gspec.Config, ethash.NewFaker(), vm.Config{})
 	defer chainman.Stop()
 	b.ReportAllocs()
 	b.ResetTimer()
@@ -283,7 +283,7 @@ func benchReadChain(b *testing.B, full bool, count uint64) {
 		if err != nil {
 			b.Fatalf("error opening database at %v: %v", dir, err)
 		}
-		chain, err := NewBlockChain(db, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
+		chain, err := NewBlockChain(db, nil, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
 		if err != nil {
 			b.Fatalf("error creating chain: %v", err)
 		}

core/block_validator.go

@@ -50,11 +50,14 @@ func NewBlockValidator(config *params.ChainConfig, blockchain *BlockChain, engin
 // validated at this point.
 func (v *BlockValidator) ValidateBody(block *types.Block) error {
 	// Check whether the block's known, and if not, that it's linkable
-	if v.bc.HasBlockAndState(block.Hash()) {
+	if v.bc.HasBlockAndState(block.Hash(), block.NumberU64()) {
 		return ErrKnownBlock
 	}
-	if !v.bc.HasBlockAndState(block.ParentHash()) {
-		return consensus.ErrUnknownAncestor
+	if !v.bc.HasBlockAndState(block.ParentHash(), block.NumberU64()-1) {
+		if !v.bc.HasBlock(block.ParentHash(), block.NumberU64()-1) {
+			return consensus.ErrUnknownAncestor
+		}
+		return consensus.ErrPrunedAncestor
 	}
 	// Header validity is known at this point, check the uncles and transactions
 	header := block.Header()

core/block_validator_test.go

@@ -42,7 +42,7 @@ func TestHeaderVerification(t *testing.T) {
 		headers[i] = block.Header()
 	}
 	// Run the header checker for blocks one-by-one, checking for both valid and invalid nonces
-	chain, _ := NewBlockChain(testdb, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
+	chain, _ := NewBlockChain(testdb, nil, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
 	defer chain.Stop()
 
 	for i := 0; i < len(blocks); i++ {
@@ -106,11 +106,11 @@ func testHeaderConcurrentVerification(t *testing.T, threads int) {
 		var results <-chan error
 
 		if valid {
-			chain, _ := NewBlockChain(testdb, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
+			chain, _ := NewBlockChain(testdb, nil, params.TestChainConfig, ethash.NewFaker(), vm.Config{})
 			_, results = chain.engine.VerifyHeaders(chain, headers, seals)
 			chain.Stop()
 		} else {
-			chain, _ := NewBlockChain(testdb, params.TestChainConfig, ethash.NewFakeFailer(uint64(len(headers)-1)), vm.Config{})
+			chain, _ := NewBlockChain(testdb, nil, params.TestChainConfig, ethash.NewFakeFailer(uint64(len(headers)-1)), vm.Config{})
 			_, results = chain.engine.VerifyHeaders(chain, headers, seals)
 			chain.Stop()
 		}
@@ -173,7 +173,7 @@ func testHeaderConcurrentAbortion(t *testing.T, threads int) {
 	defer runtime.GOMAXPROCS(old)
 
 	// Start the verifications and immediately abort
-	chain, _ := NewBlockChain(testdb, params.TestChainConfig, ethash.NewFakeDelayer(time.Millisecond), vm.Config{})
+	chain, _ := NewBlockChain(testdb, nil, params.TestChainConfig, ethash.NewFakeDelayer(time.Millisecond), vm.Config{})
 	defer chain.Stop()
 
 	abort, results := chain.engine.VerifyHeaders(chain, headers, seals)