plugeth/swarm/network/stream/delivery.go

289 lines
7.9 KiB
Go
Raw Normal View History

2018-06-20 12:06:27 +00:00
// Copyright 2018 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package stream
import (
"context"
2018-06-20 12:06:27 +00:00
"errors"
"fmt"
2018-06-20 12:06:27 +00:00
"github.com/ethereum/go-ethereum/metrics"
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
"github.com/ethereum/go-ethereum/p2p/enode"
2018-06-20 12:06:27 +00:00
"github.com/ethereum/go-ethereum/swarm/log"
"github.com/ethereum/go-ethereum/swarm/network"
"github.com/ethereum/go-ethereum/swarm/spancontext"
2018-06-20 12:06:27 +00:00
"github.com/ethereum/go-ethereum/swarm/storage"
opentracing "github.com/opentracing/opentracing-go"
2018-06-20 12:06:27 +00:00
)
const (
swarmChunkServerStreamName = "RETRIEVE_REQUEST"
deliveryCap = 32
)
var (
processReceivedChunksCount = metrics.NewRegisteredCounter("network.stream.received_chunks.count", nil)
handleRetrieveRequestMsgCount = metrics.NewRegisteredCounter("network.stream.handle_retrieve_request_msg.count", nil)
retrieveChunkFail = metrics.NewRegisteredCounter("network.stream.retrieve_chunks_fail.count", nil)
2018-06-20 12:06:27 +00:00
requestFromPeersCount = metrics.NewRegisteredCounter("network.stream.request_from_peers.count", nil)
requestFromPeersEachCount = metrics.NewRegisteredCounter("network.stream.request_from_peers_each.count", nil)
)
type Delivery struct {
chunkStore storage.SyncChunkStore
kad *network.Kademlia
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
getPeer func(enode.ID) *Peer
2018-06-20 12:06:27 +00:00
}
func NewDelivery(kad *network.Kademlia, chunkStore storage.SyncChunkStore) *Delivery {
return &Delivery{
chunkStore: chunkStore,
kad: kad,
2018-06-20 12:06:27 +00:00
}
}
// SwarmChunkServer implements Server
type SwarmChunkServer struct {
deliveryC chan []byte
batchC chan []byte
chunkStore storage.ChunkStore
2018-06-20 12:06:27 +00:00
currentLen uint64
quit chan struct{}
}
// NewSwarmChunkServer is SwarmChunkServer constructor
func NewSwarmChunkServer(chunkStore storage.ChunkStore) *SwarmChunkServer {
2018-06-20 12:06:27 +00:00
s := &SwarmChunkServer{
deliveryC: make(chan []byte, deliveryCap),
batchC: make(chan []byte),
chunkStore: chunkStore,
quit: make(chan struct{}),
2018-06-20 12:06:27 +00:00
}
go s.processDeliveries()
return s
}
// processDeliveries handles delivered chunk hashes
func (s *SwarmChunkServer) processDeliveries() {
var hashes []byte
var batchC chan []byte
for {
select {
case <-s.quit:
return
case hash := <-s.deliveryC:
hashes = append(hashes, hash...)
batchC = s.batchC
case batchC <- hashes:
hashes = nil
batchC = nil
}
}
}
// SessionIndex returns zero in all cases for SwarmChunkServer.
func (s *SwarmChunkServer) SessionIndex() (uint64, error) {
return 0, nil
}
2018-06-20 12:06:27 +00:00
// SetNextBatch
func (s *SwarmChunkServer) SetNextBatch(_, _ uint64) (hashes []byte, from uint64, to uint64, proof *HandoverProof, err error) {
select {
case hashes = <-s.batchC:
case <-s.quit:
return
}
from = s.currentLen
s.currentLen += uint64(len(hashes))
to = s.currentLen
return
}
// Close needs to be called on a stream server
func (s *SwarmChunkServer) Close() {
close(s.quit)
}
// GetData retrives chunk data from db store
func (s *SwarmChunkServer) GetData(ctx context.Context, key []byte) ([]byte, error) {
chunk, err := s.chunkStore.Get(ctx, storage.Address(key))
if err != nil {
2018-06-20 12:06:27 +00:00
return nil, err
}
return chunk.Data(), nil
2018-06-20 12:06:27 +00:00
}
// RetrieveRequestMsg is the protocol msg for chunk retrieve requests
type RetrieveRequestMsg struct {
Addr storage.Address
SkipCheck bool
HopCount uint8
2018-06-20 12:06:27 +00:00
}
func (d *Delivery) handleRetrieveRequestMsg(ctx context.Context, sp *Peer, req *RetrieveRequestMsg) error {
2018-06-20 12:06:27 +00:00
log.Trace("received request", "peer", sp.ID(), "hash", req.Addr)
handleRetrieveRequestMsgCount.Inc(1)
var osp opentracing.Span
ctx, osp = spancontext.StartSpan(
ctx,
"retrieve.request")
defer osp.Finish()
s, err := sp.getServer(NewStream(swarmChunkServerStreamName, "", true))
2018-06-20 12:06:27 +00:00
if err != nil {
return err
}
streamer := s.Server.(*SwarmChunkServer)
var cancel func()
// TODO: do something with this hardcoded timeout, maybe use TTL in the future
ctx = context.WithValue(ctx, "peer", sp.ID().String())
ctx = context.WithValue(ctx, "hopcount", req.HopCount)
ctx, cancel = context.WithTimeout(ctx, network.RequestTimeout)
go func() {
select {
case <-ctx.Done():
case <-streamer.quit:
2018-06-20 12:06:27 +00:00
}
cancel()
}()
go func() {
chunk, err := d.chunkStore.Get(ctx, req.Addr)
if err != nil {
retrieveChunkFail.Inc(1)
log.Debug("ChunkStore.Get can not retrieve chunk", "peer", sp.ID().String(), "addr", req.Addr, "hopcount", req.HopCount, "err", err)
return
}
if req.SkipCheck {
syncing := false
err = sp.Deliver(ctx, chunk, s.priority, syncing)
if err != nil {
log.Warn("ERROR in handleRetrieveRequestMsg", "err", err)
2018-06-20 12:06:27 +00:00
}
return
2018-06-20 12:06:27 +00:00
}
select {
case streamer.deliveryC <- chunk.Address()[:]:
case <-streamer.quit:
}
}()
2018-06-20 12:06:27 +00:00
return nil
}
//Chunk delivery always uses the same message type....
2018-06-20 12:06:27 +00:00
type ChunkDeliveryMsg struct {
Addr storage.Address
SData []byte // the stored chunk Data (incl size)
peer *Peer // set in handleChunkDeliveryMsg
}
//...but swap accounting needs to disambiguate if it is a delivery for syncing or for retrieval
//as it decides based on message type if it needs to account for this message or not
//defines a chunk delivery for retrieval (with accounting)
type ChunkDeliveryMsgRetrieval ChunkDeliveryMsg
//defines a chunk delivery for syncing (without accounting)
type ChunkDeliveryMsgSyncing ChunkDeliveryMsg
// TODO: Fix context SNAFU
func (d *Delivery) handleChunkDeliveryMsg(ctx context.Context, sp *Peer, req *ChunkDeliveryMsg) error {
var osp opentracing.Span
ctx, osp = spancontext.StartSpan(
ctx,
"chunk.delivery")
defer osp.Finish()
processReceivedChunksCount.Inc(1)
2018-06-20 12:06:27 +00:00
go func() {
req.peer = sp
err := d.chunkStore.Put(ctx, storage.NewChunk(req.Addr, req.SData))
if err != nil {
2018-06-20 12:06:27 +00:00
if err == storage.ErrChunkInvalid {
// we removed this log because it spams the logs
// TODO: Enable this log line
// log.Warn("invalid chunk delivered", "peer", sp.ID(), "chunk", req.Addr, )
2018-06-20 12:06:27 +00:00
req.peer.Drop(err)
}
}
}()
return nil
2018-06-20 12:06:27 +00:00
}
// RequestFromPeers sends a chunk retrieve request to
all: new p2p node representation (#17643) Package p2p/enode provides a generalized representation of p2p nodes which can contain arbitrary information in key/value pairs. It is also the new home for the node database. The "v4" identity scheme is also moved here from p2p/enr to remove the dependency on Ethereum crypto from that package. Record signature handling is changed significantly. The identity scheme registry is removed and acceptable schemes must be passed to any method that needs identity. This means records must now be validated explicitly after decoding. The enode API is designed to make signature handling easy and safe: most APIs around the codebase work with enode.Node, which is a wrapper around a valid record. Going from enr.Record to enode.Node requires a valid signature. * p2p/discover: port to p2p/enode This ports the discovery code to the new node representation in p2p/enode. The wire protocol is unchanged, this can be considered a refactoring change. The Kademlia table can now deal with nodes using an arbitrary identity scheme. This requires a few incompatible API changes: - Table.Lookup is not available anymore. It used to take a public key as argument because v4 protocol requires one. Its replacement is LookupRandom. - Table.Resolve takes *enode.Node instead of NodeID. This is also for v4 protocol compatibility because nodes cannot be looked up by ID alone. - Types Node and NodeID are gone. Further commits in the series will be fixes all over the the codebase to deal with those removals. * p2p: port to p2p/enode and discovery changes This adapts package p2p to the changes in p2p/discover. All uses of discover.Node and discover.NodeID are replaced by their equivalents from p2p/enode. New API is added to retrieve the enode.Node instance of a peer. The behavior of Server.Self with discovery disabled is improved. It now tries much harder to report a working IP address, falling back to 127.0.0.1 if no suitable address can be determined through other means. These changes were needed for tests of other packages later in the series. * p2p/simulations, p2p/testing: port to p2p/enode No surprises here, mostly replacements of discover.Node, discover.NodeID with their new equivalents. The 'interesting' API changes are: - testing.ProtocolSession tracks complete nodes, not just their IDs. - adapters.NodeConfig has a new method to create a complete node. These changes were needed to make swarm tests work. Note that the NodeID change makes the code incompatible with old simulation snapshots. * whisper/whisperv5, whisper/whisperv6: port to p2p/enode This port was easy because whisper uses []byte for node IDs and URL strings in the API. * eth: port to p2p/enode Again, easy to port because eth uses strings for node IDs and doesn't care about node information in any way. * les: port to p2p/enode Apart from replacing discover.NodeID with enode.ID, most changes are in the server pool code. It now deals with complete nodes instead of (Pubkey, IP, Port) triples. The database format is unchanged for now, but we should probably change it to use the node database later. * node: port to p2p/enode This change simply replaces discover.Node and discover.NodeID with their new equivalents. * swarm/network: port to p2p/enode Swarm has its own node address representation, BzzAddr, containing both an overlay address (the hash of a secp256k1 public key) and an underlay address (enode:// URL). There are no changes to the BzzAddr format in this commit, but certain operations such as creating a BzzAddr from a node ID are now impossible because node IDs aren't public keys anymore. Most swarm-related changes in the series remove uses of NewAddrFromNodeID, replacing it with NewAddr which takes a complete node as argument. ToOverlayAddr is removed because we can just use the node ID directly.
2018-09-24 22:59:00 +00:00
func (d *Delivery) RequestFromPeers(ctx context.Context, req *network.Request) (*enode.ID, chan struct{}, error) {
2018-06-20 12:06:27 +00:00
requestFromPeersCount.Inc(1)
var sp *Peer
spID := req.Source
if spID != nil {
sp = d.getPeer(*spID)
if sp == nil {
return nil, nil, fmt.Errorf("source peer %v not found", spID.String())
}
} else {
d.kad.EachConn(req.Addr[:], 255, func(p *network.Peer, po int) bool {
id := p.ID()
if p.LightNode {
// skip light nodes
return true
}
if req.SkipPeer(id.String()) {
log.Trace("Delivery.RequestFromPeers: skip peer", "peer id", id)
2018-06-20 12:06:27 +00:00
return true
}
sp = d.getPeer(id)
// sp is nil, when we encounter a peer that is not registered for delivery, i.e. doesn't support the `stream` protocol
if sp == nil {
return true
}
// nodes that do not provide stream protocol
// should not be requested, e.g. bootnodes
if !p.HasCap("stream") {
// TODO: if we have no errors, delete this if
log.Error("Delivery.RequestFromPeers: peer doesn't have stream cap. we should have returned at sp == nil")
return true
}
spID = &id
return false
})
2018-06-20 12:06:27 +00:00
if sp == nil {
return nil, nil, errors.New("no peer found")
2018-06-20 12:06:27 +00:00
}
}
err := sp.SendPriority(ctx, &RetrieveRequestMsg{
Addr: req.Addr,
SkipCheck: req.SkipCheck,
HopCount: req.HopCount,
}, Top)
if err != nil {
return nil, nil, err
}
requestFromPeersEachCount.Inc(1)
return spID, sp.quit, nil
2018-06-20 12:06:27 +00:00
}