cmd/geth: add prune history command

cmd/geth/chaincmd.go

@@ -35,6 +35,7 @@ import (
 	"github.com/ethereum/go-ethereum/core/state"
 	"github.com/ethereum/go-ethereum/core/types"
 	"github.com/ethereum/go-ethereum/crypto"
+	"github.com/ethereum/go-ethereum/eth/ethconfig"
 	"github.com/ethereum/go-ethereum/ethdb"
 	"github.com/ethereum/go-ethereum/internal/era"
 	"github.com/ethereum/go-ethereum/log"
@@ -189,6 +190,18 @@ It's deprecated, please use "geth db import" instead.
 This command dumps out the state for a given block (or latest, if none provided).
 `,
 	}
+
+	pruneCommand = &cli.Command{
+		Action:    pruneHistory,
+		Name:      "prune-history",
+		Usage:     "Prune blockchain history (block bodies and receipts) up to the merge block",
+		ArgsUsage: "",
+		Flags:     utils.DatabaseFlags,
+		Description: `
+The prune-history command removes historical block bodies and receipts from the
+blockchain database up to the merge block, while preserving block headers. This
+helps reduce storage requirements for nodes that don't need full historical data.`,
+	}
 )
 
 // initGenesis will initialise the given JSON format genesis file and writes it as
@@ -598,3 +611,73 @@ func hashish(x string) bool {
 	_, err := strconv.Atoi(x)
 	return err != nil
 }
+
+func pruneHistory(ctx *cli.Context) error {
+	stack, _ := makeConfigNode(ctx)
+	defer stack.Close()
+
+	// Open the chain database
+	chain, chaindb := utils.MakeChain(ctx, stack, false)
+	defer chaindb.Close()
+	defer chain.Stop()
+
+	// Pruning only supported for mainnet and sepolia.
+	if chain.Config().ChainID.Cmp(params.MainnetChainConfig.ChainID) != 0 && chain.Config().ChainID.Cmp(params.SepoliaChainConfig.ChainID) != 0 {
+		log.Info("Chain pruning not supported for this network")
+		return nil
+	}
+
+	// Determine the prune point. This will be the first PoS block.
+	prunePoint, ok := ethconfig.HistoryPrunePoints[chain.Genesis().Hash()]
+	if !ok || prunePoint == nil {
+		return errors.New("prune point not found")
+	}
+	var (
+		mergeBlock     = prunePoint.BlockNumber
+		mergeBlockHash = prunePoint.BlockHash.Hex()
+	)
+
+	// Check we're far enough past merge to ensure all data is in freezer
+	currentHeader := chain.CurrentHeader()
+	if currentHeader == nil {
+		return errors.New("current header not found")
+	}
+	if currentHeader.Number.Uint64() < mergeBlock+params.FullImmutabilityThreshold {
+		return fmt.Errorf("chain not far enough past merge block, need %d more blocks",
+			mergeBlock+params.FullImmutabilityThreshold-currentHeader.Number.Uint64())
+	}
+
+	// Double-check the prune block in db has the expected hash.
+	hash := rawdb.ReadCanonicalHash(chaindb, mergeBlock)
+	if hash != common.HexToHash(mergeBlockHash) {
+		return fmt.Errorf("merge block hash mismatch: got %s, want %s", hash.Hex(), mergeBlockHash)
+	}
+
+	txlookupTail := rawdb.ReadTxIndexTail(chaindb)
+
+	log.Info("Starting chain pruning",
+		"currentHeight", currentHeader.Number,
+		"mergeBlock", mergeBlock,
+		"mergeBlockHash", mergeBlockHash)
+
+	start := time.Now()
+
+	// First prune the transaction lookup index as
+	// it requires the block bodies.
+	if txlookupTail != nil && *txlookupTail < mergeBlock {
+		rawdb.UnindexTransactions(chaindb, *txlookupTail, mergeBlock, nil, false)
+	}
+
+	// TODO(s1na): what if there is a crash between the two prune operations?
+
+	// Truncate everything up to merge block
+	if _, err := chaindb.TruncateTail(mergeBlock); err != nil {
+		return fmt.Errorf("failed to truncate ancient data: %v", err)
+	}
+
+	log.Info("Chain pruning completed",
+		"prunedUpTo", mergeBlock,
+		"elapsed", common.PrettyDuration(time.Since(start)))
+
+	return nil
+}

cmd/geth/main.go

@@ -226,6 +226,7 @@ func init() {
 		removedbCommand,
 		dumpCommand,
 		dumpGenesisCommand,
+		pruneCommand,
 		// See accountcmd.go:
 		accountCommand,
 		walletCommand,

core/rawdb/database.go

@@ -222,6 +222,14 @@ func NewDatabaseWithFreezer(db ethdb.KeyValueStore, ancient string, namespace st
 	// it to the freezer content.
 	if kvgenesis, _ := db.Get(headerHashKey(0)); len(kvgenesis) > 0 {
 		if frozen, _ := frdb.Ancients(); frozen > 0 {
+			tail, err := frdb.Tail()
+			if err != nil {
+				printChainMetadata(db)
+				return nil, fmt.Errorf("failed to retrieve tail from freezer %v", err)
+			}
+			// If tail > 0, the history has been pruned and genesis block is not anymore in the freezer.
+			// TODO: we need another way to verify the network for kvdb and freezer match.
+			if tail == 0 {
 				// If the freezer already contains something, ensure that the genesis blocks
 				// match, otherwise we might mix up freezers across chains and destroy both
 				// the freezer and the key-value store.
@@ -233,6 +241,7 @@ func NewDatabaseWithFreezer(db ethdb.KeyValueStore, ancient string, namespace st
 					printChainMetadata(db)
 					return nil, fmt.Errorf("genesis mismatch: %#x (leveldb) != %#x (ancients)", kvgenesis, frgenesis)
 				}
+			}
 			// Key-value store and freezer belong to the same network. Ensure that they
 			// are contiguous, otherwise we might end up with a non-functional freezer.
 			if kvhash, _ := db.Get(headerHashKey(frozen)); len(kvhash) == 0 {