prysm-pulse/beacon-chain/sync/initial-sync/service.go
Nishant Das 7029ed109d
Prune Outdated Deposits after Chainstart (#2090)
* disable pending deposits

* Revert "disable pending deposits"

This reverts commit 1525e01602a70737f8688cdbd63a5ce2ab721d92.

* adding new methods

* gazelle and other fixes

* spacing

* spacing

* Update beacon-chain/sync/initial-sync/service.go

Co-Authored-By: nisdas <nish1993@hotmail.com>

* remove comment
2019-03-28 00:15:29 +08:00

614 lines
20 KiB
Go

// Package initialsync is run by the beacon node when the local chain is
// behind the network's longest chain. Initial sync works as follows:
// The node requests for the slot number of the most recent finalized block.
// The node then builds from the most recent finalized block by requesting for subsequent
// blocks by slot number. Once the service detects that the local chain is caught up with
// the network, the service hands over control to the regular sync service.
// Note: The behavior of initialsync will likely change as the specification changes.
// The most significant and highly probable change will be determining where to sync from.
// The beacon chain may sync from a block in the pasts X months in order to combat long-range attacks
// (see here: https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#what-is-weak-subjectivity)
package initialsync
import (
"context"
"errors"
"fmt"
"math/big"
"runtime/debug"
"strings"
"sync"
"time"
"github.com/ethereum/go-ethereum/common"
"github.com/gogo/protobuf/proto"
peer "github.com/libp2p/go-libp2p-peer"
"github.com/prysmaticlabs/prysm/beacon-chain/db"
pb "github.com/prysmaticlabs/prysm/proto/beacon/p2p/v1"
"github.com/prysmaticlabs/prysm/shared/bytesutil"
"github.com/prysmaticlabs/prysm/shared/event"
"github.com/prysmaticlabs/prysm/shared/hashutil"
"github.com/prysmaticlabs/prysm/shared/p2p"
"github.com/prysmaticlabs/prysm/shared/params"
"github.com/sirupsen/logrus"
"go.opencensus.io/trace"
)
var log = logrus.WithField("prefix", "initial-sync")
var debugError = "debug:"
// Config defines the configurable properties of InitialSync.
//
type Config struct {
SyncPollingInterval time.Duration
BlockBufferSize int
BlockAnnounceBufferSize int
BatchedBlockBufferSize int
StateBufferSize int
BeaconDB *db.BeaconDB
P2P p2pAPI
SyncService syncService
ChainService chainService
PowChain powChainService
}
// DefaultConfig provides the default configuration for a sync service.
// SyncPollingInterval determines how frequently the service checks that initial sync is complete.
// BlockBufferSize determines that buffer size of the `blockBuf` channel.
// StateBufferSize determines the buffer size of the `stateBuf` channel.
func DefaultConfig() *Config {
return &Config{
SyncPollingInterval: time.Duration(params.BeaconConfig().SyncPollingInterval) * time.Second,
BlockBufferSize: params.BeaconConfig().DefaultBufferSize,
BatchedBlockBufferSize: params.BeaconConfig().DefaultBufferSize,
BlockAnnounceBufferSize: params.BeaconConfig().DefaultBufferSize,
StateBufferSize: params.BeaconConfig().DefaultBufferSize,
}
}
type p2pAPI interface {
p2p.Broadcaster
p2p.Sender
Subscribe(msg proto.Message, channel chan p2p.Message) event.Subscription
}
type chainService interface {
ReceiveBlock(ctx context.Context, block *pb.BeaconBlock) (*pb.BeaconState, error)
ApplyForkChoiceRule(ctx context.Context, block *pb.BeaconBlock, computedState *pb.BeaconState) error
}
type powChainService interface {
BlockExists(ctx context.Context, hash common.Hash) (bool, *big.Int, error)
}
// SyncService is the interface for the Sync service.
// InitialSync calls `Start` when initial sync completes.
type syncService interface {
Start()
ResumeSync()
}
// InitialSync defines the main class in this package.
// See the package comments for a general description of the service's functions.
type InitialSync struct {
ctx context.Context
cancel context.CancelFunc
p2p p2pAPI
syncService syncService
chainService chainService
db *db.BeaconDB
powchain powChainService
blockAnnounceBuf chan p2p.Message
batchedBlockBuf chan p2p.Message
blockBuf chan p2p.Message
stateBuf chan p2p.Message
currentSlot uint64
highestObservedSlot uint64
beaconStateSlot uint64
syncPollingInterval time.Duration
inMemoryBlocks map[uint64]*pb.BeaconBlock
syncedFeed *event.Feed
reqState bool
stateRootOfHighestObservedSlot [32]byte
mutex *sync.Mutex
}
// NewInitialSyncService constructs a new InitialSyncService.
// This method is normally called by the main node.
func NewInitialSyncService(ctx context.Context,
cfg *Config,
) *InitialSync {
ctx, cancel := context.WithCancel(ctx)
blockBuf := make(chan p2p.Message, cfg.BlockBufferSize)
stateBuf := make(chan p2p.Message, cfg.StateBufferSize)
blockAnnounceBuf := make(chan p2p.Message, cfg.BlockAnnounceBufferSize)
batchedBlockBuf := make(chan p2p.Message, cfg.BatchedBlockBufferSize)
return &InitialSync{
ctx: ctx,
cancel: cancel,
p2p: cfg.P2P,
syncService: cfg.SyncService,
chainService: cfg.ChainService,
db: cfg.BeaconDB,
powchain: cfg.PowChain,
currentSlot: params.BeaconConfig().GenesisSlot,
highestObservedSlot: params.BeaconConfig().GenesisSlot,
beaconStateSlot: params.BeaconConfig().GenesisSlot,
blockBuf: blockBuf,
stateBuf: stateBuf,
batchedBlockBuf: batchedBlockBuf,
blockAnnounceBuf: blockAnnounceBuf,
syncPollingInterval: cfg.SyncPollingInterval,
inMemoryBlocks: map[uint64]*pb.BeaconBlock{},
syncedFeed: new(event.Feed),
reqState: false,
stateRootOfHighestObservedSlot: [32]byte{},
mutex: new(sync.Mutex),
}
}
// Start begins the goroutine.
func (s *InitialSync) Start() {
cHead, err := s.db.ChainHead()
if err != nil {
log.Errorf("Unable to get chain head %v", err)
}
s.currentSlot = cHead.Slot
var reqState bool
// setting genesis bool
if cHead.Slot == params.BeaconConfig().GenesisSlot || s.isSlotDiffLarge() {
reqState = true
}
s.reqState = reqState
go func() {
ticker := time.NewTicker(s.syncPollingInterval)
s.run(ticker.C)
ticker.Stop()
}()
go s.checkInMemoryBlocks()
}
// Stop kills the initial sync goroutine.
func (s *InitialSync) Stop() error {
log.Info("Stopping service")
s.cancel()
return nil
}
// InitializeObservedSlot sets the highest observed slot.
func (s *InitialSync) InitializeObservedSlot(slot uint64) {
s.highestObservedSlot = slot
}
// InitializeStateRoot sets the state root of the highest observed slot.
func (s *InitialSync) InitializeStateRoot(root [32]byte) {
s.stateRootOfHighestObservedSlot = root
}
// SyncedFeed returns a feed which fires a message once the node is synced
func (s *InitialSync) SyncedFeed() *event.Feed {
return s.syncedFeed
}
// run is the main goroutine for the initial sync service.
// delayChan is explicitly passed into this function to facilitate tests that don't require a timeout.
// It is assumed that the goroutine `run` is only called once per instance.
func (s *InitialSync) run(delayChan <-chan time.Time) {
blockSub := s.p2p.Subscribe(&pb.BeaconBlockResponse{}, s.blockBuf)
batchedBlocksub := s.p2p.Subscribe(&pb.BatchedBeaconBlockResponse{}, s.batchedBlockBuf)
blockAnnounceSub := s.p2p.Subscribe(&pb.BeaconBlockAnnounce{}, s.blockAnnounceBuf)
beaconStateSub := s.p2p.Subscribe(&pb.BeaconStateResponse{}, s.stateBuf)
defer func() {
blockSub.Unsubscribe()
blockAnnounceSub.Unsubscribe()
beaconStateSub.Unsubscribe()
batchedBlocksub.Unsubscribe()
close(s.batchedBlockBuf)
close(s.blockBuf)
close(s.stateBuf)
}()
if s.reqState {
if err := s.requestStateFromPeer(s.ctx, s.stateRootOfHighestObservedSlot[:], p2p.AnyPeer); err != nil {
log.Errorf("Could not request state from peer %v", err)
}
} else {
// Send out a batch request
s.requestBatchedBlocks(s.currentSlot+1, s.highestObservedSlot)
}
for {
select {
case <-s.ctx.Done():
log.Debug("Exiting goroutine")
return
case <-delayChan:
if s.checkSyncStatus() {
return
}
case msg := <-s.blockAnnounceBuf:
safelyHandleMessage(s.processBlockAnnounce, msg)
case msg := <-s.blockBuf:
safelyHandleMessage(func(message p2p.Message) {
data := message.Data.(*pb.BeaconBlockResponse)
s.processBlock(message.Ctx, data.Block, message.Peer)
}, msg)
case msg := <-s.stateBuf:
safelyHandleMessage(s.processState, msg)
case msg := <-s.batchedBlockBuf:
safelyHandleMessage(s.processBatchedBlocks, msg)
}
}
}
// safelyHandleMessage will recover and log any panic that occurs from the
// function argument.
func safelyHandleMessage(fn func(p2p.Message), msg p2p.Message) {
defer func() {
if r := recover(); r != nil {
printedMsg := "message contains no data"
if msg.Data != nil {
printedMsg = proto.MarshalTextString(msg.Data)
}
log.WithFields(logrus.Fields{
"r": r,
"msg": printedMsg,
}).Error("Panicked when handling p2p message! Recovering...")
debug.PrintStack()
if msg.Ctx == nil {
return
}
if span := trace.FromContext(msg.Ctx); span != nil {
span.SetStatus(trace.Status{
Code: trace.StatusCodeInternal,
Message: fmt.Sprintf("Panic: %v", r),
})
}
}
}()
// Fingers crossed that it doesn't panic...
fn(msg)
}
// checkInMemoryBlocks is another routine which will run concurrently with the
// main routine for initial sync, where it checks the blocks saved in memory regularly
// to see if the blocks are valid enough to be processed.
func (s *InitialSync) checkInMemoryBlocks() {
for {
select {
case <-s.ctx.Done():
return
default:
if s.currentSlot == s.highestObservedSlot {
return
}
s.mutex.Lock()
if block, ok := s.inMemoryBlocks[s.currentSlot+1]; ok && s.currentSlot+1 <= s.highestObservedSlot {
s.processBlock(s.ctx, block, p2p.AnyPeer)
}
s.mutex.Unlock()
}
}
}
// checkSyncStatus verifies if the beacon node is correctly synced with its peers up to their
// latest canonical head. If not, then it requests batched blocks up to the highest observed slot.
func (s *InitialSync) checkSyncStatus() bool {
if s.reqState {
if err := s.requestStateFromPeer(s.ctx, s.stateRootOfHighestObservedSlot[:], p2p.AnyPeer); err != nil {
log.Errorf("Could not request state from peer %v", err)
}
return false
}
if s.highestObservedSlot == s.currentSlot {
log.Info("Exiting initial sync and starting normal sync")
s.syncedFeed.Send(s.currentSlot)
s.syncService.ResumeSync()
return true
}
// requests multiple blocks so as to save and sync quickly.
s.requestBatchedBlocks(s.currentSlot+1, s.highestObservedSlot)
return false
}
func (s *InitialSync) processBlockAnnounce(msg p2p.Message) {
ctx, span := trace.StartSpan(msg.Ctx, "beacon-chain.sync.initial-sync.processBlockAnnounce")
defer span.End()
data := msg.Data.(*pb.BeaconBlockAnnounce)
recBlockAnnounce.Inc()
if s.reqState {
if err := s.requestStateFromPeer(ctx, s.stateRootOfHighestObservedSlot[:], msg.Peer); err != nil {
log.Errorf("Could not request state from peer %v", err)
}
return
}
if data.SlotNumber > s.highestObservedSlot {
s.highestObservedSlot = data.SlotNumber
}
s.requestBatchedBlocks(s.currentSlot+1, s.highestObservedSlot)
log.Debugf("Successfully requested the next block with slot: %d", data.SlotNumber-params.BeaconConfig().GenesisSlot)
}
// processBlock is the main method that validates each block which is received
// for initial sync. It checks if the blocks are valid and then will continue to
// process and save it into the db.
func (s *InitialSync) processBlock(ctx context.Context, block *pb.BeaconBlock, peerID peer.ID) {
ctx, span := trace.StartSpan(ctx, "beacon-chain.sync.initial-sync.processBlock")
defer span.End()
recBlock.Inc()
if block.Slot > s.highestObservedSlot {
s.highestObservedSlot = block.Slot
}
if block.Slot < s.currentSlot {
return
}
// requesting beacon state if there is no saved state.
if s.reqState {
if err := s.requestStateFromPeer(s.ctx, s.stateRootOfHighestObservedSlot[:], peerID); err != nil {
log.Errorf("Could not request beacon state from peer: %v", err)
}
return
}
// if it isn't the block in the next slot we check if it is a skipped slot.
// if it isn't skipped we save it in memory.
if block.Slot != (s.currentSlot + 1) {
// if parent exists we validate the block.
if s.doesParentExist(block) {
if err := s.validateAndSaveNextBlock(ctx, block); err != nil {
// Debug error so as not to have noisy error logs
if strings.HasPrefix(err.Error(), debugError) {
log.Debug(strings.TrimPrefix(err.Error(), debugError))
return
}
log.Errorf("Unable to save block: %v", err)
}
return
}
s.mutex.Lock()
defer s.mutex.Unlock()
if _, ok := s.inMemoryBlocks[block.Slot]; !ok {
s.inMemoryBlocks[block.Slot] = block
}
return
}
if err := s.validateAndSaveNextBlock(ctx, block); err != nil {
// Debug error so as not to have noisy error logs
if strings.HasPrefix(err.Error(), debugError) {
log.Debug(strings.TrimPrefix(err.Error(), debugError))
return
}
log.Errorf("Unable to save block: %v", err)
}
}
// processBatchedBlocks processes all the received blocks from
// the p2p message.
func (s *InitialSync) processBatchedBlocks(msg p2p.Message) {
ctx, span := trace.StartSpan(msg.Ctx, "beacon-chain.sync.initial-sync.processBatchedBlocks")
defer span.End()
batchedBlockReq.Inc()
response := msg.Data.(*pb.BatchedBeaconBlockResponse)
batchedBlocks := response.BatchedBlocks
if len(batchedBlocks) == 0 {
// Do not process empty response
return
}
log.Debug("Processing batched block response")
for _, block := range batchedBlocks {
s.processBlock(ctx, block, msg.Peer)
}
log.Debug("Finished processing batched blocks")
}
func (s *InitialSync) processState(msg p2p.Message) {
ctx, span := trace.StartSpan(msg.Ctx, "beacon-chain.sync.initial-sync.processState")
defer span.End()
data := msg.Data.(*pb.BeaconStateResponse)
beaconState := data.BeaconState
recState.Inc()
if s.currentSlot > beaconState.FinalizedEpoch*params.BeaconConfig().SlotsPerEpoch {
return
}
if err := s.db.SaveCurrentAndFinalizedState(beaconState); err != nil {
log.Errorf("Unable to set beacon state for initial sync %v", err)
return
}
if err := s.db.SaveFinalizedBlock(beaconState.LatestBlock); err != nil {
log.Errorf("Could not save finalized block %v", err)
return
}
if err := s.db.SaveBlock(beaconState.LatestBlock); err != nil {
log.Errorf("Could not save block %v", err)
return
}
if err := s.db.UpdateChainHead(beaconState.LatestBlock, beaconState); err != nil {
log.Errorf("Could not update chainhead %v", err)
return
}
if err := s.db.SaveJustifiedState(beaconState); err != nil {
log.Errorf("Could not set beacon state for initial sync %v", err)
return
}
if err := s.db.SaveJustifiedBlock(beaconState.LatestBlock); err != nil {
log.Errorf("Could not save finalized block %v", err)
return
}
h, err := hashutil.HashProto(beaconState)
if err != nil {
log.Error(err)
return
}
exists, blkNum, err := s.powchain.BlockExists(ctx, bytesutil.ToBytes32(beaconState.LatestEth1Data.BlockHash32))
if err != nil {
log.Errorf("Unable to get powchain block %v", err)
}
if !exists {
log.Error("Latest ETH1 block doesn't exist in the pow chain")
return
}
s.db.PrunePendingDeposits(ctx, blkNum)
if h == s.stateRootOfHighestObservedSlot {
s.reqState = false
}
// sets the current slot to the last finalized slot of the
// beacon state to begin our sync from.
s.currentSlot = beaconState.FinalizedEpoch * params.BeaconConfig().SlotsPerEpoch
s.beaconStateSlot = beaconState.Slot
log.Debugf("Successfully saved beacon state with the last finalized slot: %d", beaconState.FinalizedEpoch*params.BeaconConfig().SlotsPerEpoch-params.BeaconConfig().GenesisSlot)
s.requestBatchedBlocks(s.currentSlot+1, s.highestObservedSlot)
}
// requestStateFromPeer always requests for the last finalized slot from a peer.
func (s *InitialSync) requestStateFromPeer(ctx context.Context, stateRoot []byte, peerID peer.ID) error {
ctx, span := trace.StartSpan(ctx, "beacon-chain.sync.initial-sync.requestStateFromPeer")
defer span.End()
stateReq.Inc()
log.Debugf("Successfully processed incoming block with state hash: %#x", stateRoot)
return s.p2p.Send(ctx, &pb.BeaconStateRequest{FinalizedStateRootHash32S: stateRoot}, peerID)
}
// requestNextBlock broadcasts a request for a block with the entered slotnumber.
func (s *InitialSync) requestNextBlockBySlot(ctx context.Context, slotNumber uint64) {
ctx, span := trace.StartSpan(ctx, "beacon-chain.sync.initial-sync.requestBlockBySlot")
defer span.End()
log.Debugf("Requesting block %d ", slotNumber)
blockReqSlot.Inc()
s.mutex.Lock()
defer s.mutex.Unlock()
if block, ok := s.inMemoryBlocks[slotNumber]; ok {
s.processBlock(ctx, block, p2p.AnyPeer)
return
}
s.p2p.Broadcast(ctx, &pb.BeaconBlockRequestBySlotNumber{SlotNumber: slotNumber})
}
// requestBatchedBlocks sends out a request for multiple blocks till a
// specified bound slot number.
func (s *InitialSync) requestBatchedBlocks(startSlot uint64, endSlot uint64) {
ctx, span := trace.StartSpan(context.Background(), "beacon-chain.sync.initial-sync.requestBatchedBlocks")
defer span.End()
sentBatchedBlockReq.Inc()
if startSlot > endSlot {
log.Debugf("Invalid batched request from slot %d to %d", startSlot-params.BeaconConfig().GenesisSlot, endSlot-params.BeaconConfig().GenesisSlot)
return
}
blockLimit := params.BeaconConfig().BatchBlockLimit
if startSlot+blockLimit < endSlot {
endSlot = startSlot + blockLimit
}
log.Debugf("Requesting batched blocks from slot %d to %d", startSlot-params.BeaconConfig().GenesisSlot, endSlot-params.BeaconConfig().GenesisSlot)
s.p2p.Broadcast(ctx, &pb.BatchedBeaconBlockRequest{
StartSlot: startSlot,
EndSlot: endSlot,
})
}
// validateAndSaveNextBlock will validate whether blocks received from the blockfetcher
// routine can be added to the chain.
func (s *InitialSync) validateAndSaveNextBlock(ctx context.Context, block *pb.BeaconBlock) error {
ctx, span := trace.StartSpan(ctx, "beacon-chain.sync.initial-sync.validateAndSaveNextBlock")
defer span.End()
root, err := hashutil.HashBeaconBlock(block)
if err != nil {
return err
}
if err := s.checkBlockValidity(ctx, block); err != nil {
return err
}
log.Infof("Saved block with root %#x and slot %d for initial sync", root, block.Slot)
s.currentSlot = block.Slot
s.mutex.Lock()
defer s.mutex.Unlock()
// delete block from memory.
if _, ok := s.inMemoryBlocks[block.Slot]; ok {
delete(s.inMemoryBlocks, block.Slot)
}
// since the block will not be processed by chainservice we save
// the block and do not send it to chainservice.
if s.beaconStateSlot >= block.Slot {
if err := s.db.SaveBlock(block); err != nil {
return err
}
return nil
}
// Send block to main chain service to be processed.
beaconState, err := s.chainService.ReceiveBlock(ctx, block)
if err != nil {
return fmt.Errorf("could not process beacon block: %v", err)
}
if err := s.chainService.ApplyForkChoiceRule(ctx, block, beaconState); err != nil {
return fmt.Errorf("could not apply fork choice rule: %v", err)
}
return nil
}
func (s *InitialSync) checkBlockValidity(ctx context.Context, block *pb.BeaconBlock) error {
ctx, span := trace.StartSpan(ctx, "beacon-chain.sync.initial-sync.checkBlockValidity")
defer span.End()
blockRoot, err := hashutil.HashBeaconBlock(block)
if err != nil {
return fmt.Errorf("could not tree hash received block: %v", err)
}
if s.db.HasBlock(blockRoot) {
return errors.New(debugError + "received a block that already exists. Exiting")
}
beaconState, err := s.db.State(ctx)
if err != nil {
return fmt.Errorf("failed to get beacon state: %v", err)
}
if block.Slot < beaconState.FinalizedEpoch*params.BeaconConfig().SlotsPerEpoch {
return errors.New(debugError + "discarding received block with a slot number smaller than the last finalized slot")
}
// Attestation from proposer not verified as, other nodes only store blocks not proposer
// attestations.
return nil
}
// isSlotDiff large checks if the difference between the current slot and highest observed
// slot isnt too large.
func (s *InitialSync) isSlotDiffLarge() bool {
slotsPerEpoch := params.BeaconConfig().SlotsPerEpoch
epochLimit := params.BeaconConfig().SyncEpochLimit
return s.currentSlot+slotsPerEpoch*epochLimit < s.highestObservedSlot
}
func (s *InitialSync) doesParentExist(block *pb.BeaconBlock) bool {
parentHash := bytesutil.ToBytes32(block.ParentRootHash32)
return s.db.HasBlock(parentHash)
}