plugeth/eth/sync.go
Felix Lange 6c73a59806 eth: limit number of sent transactions based on message size
Nodes that are out of sync will queue many transactions, which causes
the initial transactions message to grow very large. Larger transactions
messages can make communication impossible if the message is too big to
send. Big transactions messages also exhaust egress bandwidth, which
degrades other peer connections.

The new approach to combat these issues is to send transactions in
smaller batches. This commit introduces a new goroutine that handles
delivery of all initial transaction transfers. Size-limited packs of
transactions are sent to one peer at a time, conserving precious egress
bandwidth.
2015-06-09 17:07:10 +02:00

346 lines
10 KiB
Go

package eth
import (
"math"
"math/rand"
"sync/atomic"
"time"
"github.com/ethereum/go-ethereum/common"
"github.com/ethereum/go-ethereum/core/types"
"github.com/ethereum/go-ethereum/eth/downloader"
"github.com/ethereum/go-ethereum/logger"
"github.com/ethereum/go-ethereum/logger/glog"
"github.com/ethereum/go-ethereum/p2p/discover"
)
const (
forceSyncCycle = 10 * time.Second // Time interval to force syncs, even if few peers are available
blockProcCycle = 500 * time.Millisecond // Time interval to check for new blocks to process
notifyCheckCycle = 100 * time.Millisecond // Time interval to allow hash notifies to fulfill before hard fetching
notifyArriveTimeout = 500 * time.Millisecond // Time allowance before an announced block is explicitly requested
notifyFetchTimeout = 5 * time.Second // Maximum alloted time to return an explicitly requested block
minDesiredPeerCount = 5 // Amount of peers desired to start syncing
blockProcAmount = 256
// This is the target size for the packs of transactions sent by txsyncLoop.
// A pack can get larger than this if a single transactions exceeds this size.
txsyncPackSize = 100 * 1024
)
// blockAnnounce is the hash notification of the availability of a new block in
// the network.
type blockAnnounce struct {
hash common.Hash
peer *peer
time time.Time
}
type txsync struct {
p *peer
txs []*types.Transaction
}
// syncTransactions starts sending all currently pending transactions to the given peer.
func (pm *ProtocolManager) syncTransactions(p *peer) {
txs := pm.txpool.GetTransactions()
if len(txs) == 0 {
return
}
select {
case pm.txsyncCh <- &txsync{p, txs}:
case <-pm.quitSync:
}
}
// txsyncLoop takes care of the initial transaction sync for each new
// connection. When a new peer appears, we relay all currently pending
// transactions. In order to minimise egress bandwidth usage, we send
// the transactions in small packs to one peer at a time.
func (pm *ProtocolManager) txsyncLoop() {
var (
pending = make(map[discover.NodeID]*txsync)
sending = false // whether a send is active
pack = new(txsync) // the pack that is being sent
done = make(chan error, 1) // result of the send
)
// send starts a sending a pack of transactions from the sync.
send := func(s *txsync) {
// Fill pack with transactions up to the target size.
size := common.StorageSize(0)
pack.p = s.p
pack.txs = pack.txs[:0]
for i := 0; i < len(s.txs) && size < txsyncPackSize; i++ {
pack.txs = append(pack.txs, s.txs[i])
size += s.txs[i].Size()
}
// Remove the transactions that will be sent.
s.txs = s.txs[:copy(s.txs, s.txs[len(pack.txs):])]
if len(s.txs) == 0 {
delete(pending, s.p.ID())
}
// Send the pack in the background.
glog.V(logger.Detail).Infof("%v: sending %d transactions (%v)", s.p.Peer, len(pack.txs), size)
sending = true
go func() { done <- pack.p.sendTransactions(pack.txs) }()
}
// pick chooses the next pending sync.
pick := func() *txsync {
if len(pending) == 0 {
return nil
}
n := rand.Intn(len(pending)) + 1
for _, s := range pending {
if n--; n == 0 {
return s
}
}
return nil
}
for {
select {
case s := <-pm.txsyncCh:
pending[s.p.ID()] = s
if !sending {
send(s)
}
case err := <-done:
sending = false
// Stop tracking peers that cause send failures.
if err != nil {
glog.V(logger.Debug).Infof("%v: tx send failed: %v", pack.p.Peer, err)
delete(pending, pack.p.ID())
}
// Schedule the next send.
if s := pick(); s != nil {
send(s)
}
case <-pm.quitSync:
return
}
}
}
// fetcher is responsible for collecting hash notifications, and periodically
// checking all unknown ones and individually fetching them.
func (pm *ProtocolManager) fetcher() {
announces := make(map[common.Hash]*blockAnnounce)
request := make(map[*peer][]common.Hash)
pending := make(map[common.Hash]*blockAnnounce)
cycle := time.Tick(notifyCheckCycle)
// Iterate the block fetching until a quit is requested
for {
select {
case notifications := <-pm.newHashCh:
// A batch of hashes the notified, schedule them for retrieval
glog.V(logger.Debug).Infof("Scheduling %d hash announcements from %s", len(notifications), notifications[0].peer.id)
for _, announce := range notifications {
announces[announce.hash] = announce
}
case <-cycle:
// Clean up any expired block fetches
for hash, announce := range pending {
if time.Since(announce.time) > notifyFetchTimeout {
delete(pending, hash)
}
}
// Check if any notified blocks failed to arrive
for hash, announce := range announces {
if time.Since(announce.time) > notifyArriveTimeout {
if !pm.chainman.HasBlock(hash) {
request[announce.peer] = append(request[announce.peer], hash)
pending[hash] = announce
}
delete(announces, hash)
}
}
if len(request) == 0 {
break
}
// Send out all block requests
for peer, hashes := range request {
glog.V(logger.Debug).Infof("Explicitly fetching %d blocks from %s", len(hashes), peer.id)
peer.requestBlocks(hashes)
}
request = make(map[*peer][]common.Hash)
case filter := <-pm.newBlockCh:
// Blocks arrived, extract any explicit fetches, return all else
var blocks types.Blocks
select {
case blocks = <-filter:
case <-pm.quitSync:
return
}
explicit, download := []*types.Block{}, []*types.Block{}
for _, block := range blocks {
hash := block.Hash()
// Filter explicitly requested blocks from hash announcements
if _, ok := pending[hash]; ok {
// Discard if already imported by other means
if !pm.chainman.HasBlock(hash) {
explicit = append(explicit, block)
} else {
delete(pending, hash)
}
} else {
download = append(download, block)
}
}
select {
case filter <- download:
case <-pm.quitSync:
return
}
// If any explicit fetches were replied to, import them
if count := len(explicit); count > 0 {
glog.V(logger.Debug).Infof("Importing %d explicitly fetched blocks", count)
// Create a closure with the retrieved blocks and origin peers
peers := make([]*peer, 0, count)
blocks := make([]*types.Block, 0, count)
for _, block := range explicit {
hash := block.Hash()
if announce := pending[hash]; announce != nil {
peers = append(peers, announce.peer)
blocks = append(blocks, block)
delete(pending, hash)
}
}
// Run the importer on a new thread
go func() {
for i := 0; i < len(blocks); i++ {
if err := pm.importBlock(peers[i], blocks[i], nil); err != nil {
glog.V(logger.Detail).Infof("Failed to import explicitly fetched block: %v", err)
return
}
}
}()
}
case <-pm.quitSync:
return
}
}
}
// syncer is responsible for periodically synchronising with the network, both
// downloading hashes and blocks as well as retrieving cached ones.
func (pm *ProtocolManager) syncer() {
forceSync := time.Tick(forceSyncCycle)
blockProc := time.Tick(blockProcCycle)
blockProcPend := int32(0)
for {
select {
case <-pm.newPeerCh:
// Make sure we have peers to select from, then sync
if pm.peers.Len() < minDesiredPeerCount {
break
}
go pm.synchronise(pm.peers.BestPeer())
case <-forceSync:
// Force a sync even if not enough peers are present
go pm.synchronise(pm.peers.BestPeer())
case <-blockProc:
// Try to pull some blocks from the downloaded
if atomic.CompareAndSwapInt32(&blockProcPend, 0, 1) {
go func() {
pm.processBlocks()
atomic.StoreInt32(&blockProcPend, 0)
}()
}
case <-pm.quitSync:
return
}
}
}
// processBlocks retrieves downloaded blocks from the download cache and tries
// to construct the local block chain with it. Note, since the block retrieval
// order matters, access to this function *must* be synchronized/serialized.
func (pm *ProtocolManager) processBlocks() error {
pm.wg.Add(1)
defer pm.wg.Done()
// Short circuit if no blocks are available for insertion
blocks := pm.downloader.TakeBlocks()
if len(blocks) == 0 {
return nil
}
glog.V(logger.Debug).Infof("Inserting chain with %d blocks (#%v - #%v)\n", len(blocks), blocks[0].RawBlock.Number(), blocks[len(blocks)-1].RawBlock.Number())
for len(blocks) != 0 && !pm.quit {
// Retrieve the first batch of blocks to insert
max := int(math.Min(float64(len(blocks)), float64(blockProcAmount)))
raw := make(types.Blocks, 0, max)
for _, block := range blocks[:max] {
raw = append(raw, block.RawBlock)
}
// Try to inset the blocks, drop the originating peer if there's an error
index, err := pm.chainman.InsertChain(raw)
if err != nil {
glog.V(logger.Debug).Infoln("Downloaded block import failed:", err)
pm.removePeer(blocks[index].OriginPeer)
pm.downloader.Cancel()
return err
}
blocks = blocks[max:]
}
return nil
}
// synchronise tries to sync up our local block chain with a remote peer, both
// adding various sanity checks as well as wrapping it with various log entries.
func (pm *ProtocolManager) synchronise(peer *peer) {
// Short circuit if no peers are available
if peer == nil {
return
}
// Make sure the peer's TD is higher than our own. If not drop.
if peer.Td().Cmp(pm.chainman.Td()) <= 0 {
return
}
// FIXME if we have the hash in our chain and the TD of the peer is
// much higher than ours, something is wrong with us or the peer.
// Check if the hash is on our own chain
head := peer.Head()
if pm.chainman.HasBlock(head) {
glog.V(logger.Debug).Infoln("Synchronisation canceled: head already known")
return
}
// Get the hashes from the peer (synchronously)
glog.V(logger.Detail).Infof("Attempting synchronisation: %v, 0x%x", peer.id, head)
err := pm.downloader.Synchronise(peer.id, head)
switch err {
case nil:
glog.V(logger.Detail).Infof("Synchronisation completed")
case downloader.ErrBusy:
glog.V(logger.Detail).Infof("Synchronisation already in progress")
case downloader.ErrTimeout, downloader.ErrBadPeer, downloader.ErrEmptyHashSet, downloader.ErrInvalidChain, downloader.ErrCrossCheckFailed:
glog.V(logger.Debug).Infof("Removing peer %v: %v", peer.id, err)
pm.removePeer(peer.id)
case downloader.ErrPendingQueue:
glog.V(logger.Debug).Infoln("Synchronisation aborted:", err)
default:
glog.V(logger.Warn).Infof("Synchronisation failed: %v", err)
}
}