tendermint/consensus/reactor.go
Anton Kaliaev f6539737de
new pubsub package
comment out failing consensus tests for now

rewrite rpc httpclient to use new pubsub package

import pubsub as tmpubsub, query as tmquery

make event IDs constants
EventKey -> EventTypeKey

rename EventsPubsub to PubSub

mempool does not use pubsub

rename eventsSub to pubsub

new subscribe API

fix channel size issues and consensus tests bugs

refactor rpc client

add missing discardFromChan method

add mutex

rename pubsub to eventBus

remove IsRunning from WSRPCConnection interface (not needed)

add a comment in broadcastNewRoundStepsAndVotes

rename registerEventCallbacks to broadcastNewRoundStepsAndVotes

See https://dave.cheney.net/2014/03/19/channel-axioms

stop eventBuses after reactor tests

remove unnecessary Unsubscribe

return subscribe helper function

move discardFromChan to where it is used

subscribe now returns an err

this gives us ability to refuse to subscribe if pubsub is at its max
capacity.

use context for control overflow

cache queries

handle err when subscribing in replay_test

rename testClientID to testSubscriber

extract var

set channel buffer capacity to 1 in replay_file

fix byzantine_test

unsubscribe from single event, not all events

refactor httpclient to return events to appropriate channels

return failing testReplayCrashBeforeWriteVote test

fix TestValidatorSetChanges

refactor code a bit

fix testReplayCrashBeforeWriteVote

add comment

fix TestValidatorSetChanges

fixes from Bucky's review

update comment [ci skip]

test TxEventBuffer

update changelog

fix TestValidatorSetChanges (2nd attempt)

only do wg.Done when no errors

benchmark event bus

create pubsub server inside NewEventBus

only expose config params (later if needed)

set buffer capacity to 0 so we are not testing cache

new tx event format: key = "Tx" plus a tag {"tx.hash": XYZ}

This should allow to subscribe to all transactions! or a specific one
using a query: "tm.events.type = Tx and tx.hash = '013ABF99434...'"

use TimeoutCommit instead of afterPublishEventNewBlockTimeout

TimeoutCommit is the time a node waits after committing a block, before
it goes into the next height. So it will finish everything from the last
block, but then wait a bit. The idea is this gives it time to hear more
votes from other validators, to strengthen the commit it includes in the
next block. But it also gives it time to hear about new transactions.

waitForBlockWithUpdatedVals

rewrite WAL crash tests

Task:
test that we can recover from any WAL crash.

Solution:
the old tests were relying on event hub being run in the same thread (we
were injecting the private validator's last signature).

when considering a rewrite, we considered two possible solutions: write
a "fuzzy" testing system where WAL is crashing upon receiving a new
message, or inject failures and trigger them in tests using something
like https://github.com/coreos/gofail.

remove sleep

no cs.Lock around wal.Save

test different cases (empty block, non-empty block, ...)

comments

add comments

test 4 cases: empty block, non-empty block, non-empty block with smaller part size, many blocks

fixes as per Bucky's last review

reset subscriptions on UnsubscribeAll

use a simple counter to track message for which we panicked

also, set a smaller part size for all test cases
2017-10-30 00:32:22 -05:00

1348 lines
40 KiB
Go

package consensus
import (
"bytes"
"context"
"fmt"
"reflect"
"sync"
"time"
"github.com/pkg/errors"
wire "github.com/tendermint/go-wire"
cmn "github.com/tendermint/tmlibs/common"
"github.com/tendermint/tmlibs/log"
cstypes "github.com/tendermint/tendermint/consensus/types"
"github.com/tendermint/tendermint/p2p"
sm "github.com/tendermint/tendermint/state"
"github.com/tendermint/tendermint/types"
)
const (
StateChannel = byte(0x20)
DataChannel = byte(0x21)
VoteChannel = byte(0x22)
VoteSetBitsChannel = byte(0x23)
maxConsensusMessageSize = 1048576 // 1MB; NOTE/TODO: keep in sync with types.PartSet sizes.
)
//-----------------------------------------------------------------------------
// ConsensusReactor defines a reactor for the consensus service.
type ConsensusReactor struct {
p2p.BaseReactor // BaseService + p2p.Switch
conS *ConsensusState
mtx sync.RWMutex
fastSync bool
eventBus *types.EventBus
}
// NewConsensusReactor returns a new ConsensusReactor with the given consensusState.
func NewConsensusReactor(consensusState *ConsensusState, fastSync bool) *ConsensusReactor {
conR := &ConsensusReactor{
conS: consensusState,
fastSync: fastSync,
}
conR.BaseReactor = *p2p.NewBaseReactor("ConsensusReactor", conR)
return conR
}
// OnStart implements BaseService.
func (conR *ConsensusReactor) OnStart() error {
conR.Logger.Info("ConsensusReactor ", "fastSync", conR.FastSync())
conR.BaseReactor.OnStart()
err := conR.broadcastNewRoundStepsAndVotes()
if err != nil {
return err
}
if !conR.FastSync() {
_, err := conR.conS.Start()
if err != nil {
return err
}
}
return nil
}
// OnStop implements BaseService
func (conR *ConsensusReactor) OnStop() {
conR.BaseReactor.OnStop()
conR.conS.Stop()
}
// SwitchToConsensus switches from fast_sync mode to consensus mode.
// It resets the state, turns off fast_sync, and starts the consensus state-machine
func (conR *ConsensusReactor) SwitchToConsensus(state *sm.State, blocksSynced int) {
conR.Logger.Info("SwitchToConsensus")
conR.conS.reconstructLastCommit(state)
// NOTE: The line below causes broadcastNewRoundStepRoutine() to
// broadcast a NewRoundStepMessage.
conR.conS.updateToState(state)
conR.mtx.Lock()
conR.fastSync = false
conR.mtx.Unlock()
if blocksSynced > 0 {
// dont bother with the WAL if we fast synced
conR.conS.doWALCatchup = false
}
conR.conS.Start()
}
// GetChannels implements Reactor
func (conR *ConsensusReactor) GetChannels() []*p2p.ChannelDescriptor {
// TODO optimize
return []*p2p.ChannelDescriptor{
&p2p.ChannelDescriptor{
ID: StateChannel,
Priority: 5,
SendQueueCapacity: 100,
},
&p2p.ChannelDescriptor{
ID: DataChannel, // maybe split between gossiping current block and catchup stuff
Priority: 10, // once we gossip the whole block there's nothing left to send until next height or round
SendQueueCapacity: 100,
RecvBufferCapacity: 50 * 4096,
},
&p2p.ChannelDescriptor{
ID: VoteChannel,
Priority: 5,
SendQueueCapacity: 100,
RecvBufferCapacity: 100 * 100,
},
&p2p.ChannelDescriptor{
ID: VoteSetBitsChannel,
Priority: 1,
SendQueueCapacity: 2,
RecvBufferCapacity: 1024,
},
}
}
// AddPeer implements Reactor
func (conR *ConsensusReactor) AddPeer(peer p2p.Peer) {
if !conR.IsRunning() {
return
}
// Create peerState for peer
peerState := NewPeerState(peer).SetLogger(conR.Logger)
peer.Set(types.PeerStateKey, peerState)
// Begin routines for this peer.
go conR.gossipDataRoutine(peer, peerState)
go conR.gossipVotesRoutine(peer, peerState)
go conR.queryMaj23Routine(peer, peerState)
// Send our state to peer.
// If we're fast_syncing, broadcast a RoundStepMessage later upon SwitchToConsensus().
if !conR.FastSync() {
conR.sendNewRoundStepMessages(peer)
}
}
// RemovePeer implements Reactor
func (conR *ConsensusReactor) RemovePeer(peer p2p.Peer, reason interface{}) {
if !conR.IsRunning() {
return
}
// TODO
//peer.Get(PeerStateKey).(*PeerState).Disconnect()
}
// Receive implements Reactor
// NOTE: We process these messages even when we're fast_syncing.
// Messages affect either a peer state or the consensus state.
// Peer state updates can happen in parallel, but processing of
// proposals, block parts, and votes are ordered by the receiveRoutine
// NOTE: blocks on consensus state for proposals, block parts, and votes
func (conR *ConsensusReactor) Receive(chID byte, src p2p.Peer, msgBytes []byte) {
if !conR.IsRunning() {
conR.Logger.Debug("Receive", "src", src, "chId", chID, "bytes", msgBytes)
return
}
_, msg, err := DecodeMessage(msgBytes)
if err != nil {
conR.Logger.Error("Error decoding message", "src", src, "chId", chID, "msg", msg, "err", err, "bytes", msgBytes)
// TODO punish peer?
return
}
conR.Logger.Debug("Receive", "src", src, "chId", chID, "msg", msg)
// Get peer states
ps := src.Get(types.PeerStateKey).(*PeerState)
switch chID {
case StateChannel:
switch msg := msg.(type) {
case *NewRoundStepMessage:
ps.ApplyNewRoundStepMessage(msg)
case *CommitStepMessage:
ps.ApplyCommitStepMessage(msg)
case *HasVoteMessage:
ps.ApplyHasVoteMessage(msg)
case *VoteSetMaj23Message:
cs := conR.conS
cs.mtx.Lock()
height, votes := cs.Height, cs.Votes
cs.mtx.Unlock()
if height != msg.Height {
return
}
// Peer claims to have a maj23 for some BlockID at H,R,S,
votes.SetPeerMaj23(msg.Round, msg.Type, ps.Peer.Key(), msg.BlockID)
// Respond with a VoteSetBitsMessage showing which votes we have.
// (and consequently shows which we don't have)
var ourVotes *cmn.BitArray
switch msg.Type {
case types.VoteTypePrevote:
ourVotes = votes.Prevotes(msg.Round).BitArrayByBlockID(msg.BlockID)
case types.VoteTypePrecommit:
ourVotes = votes.Precommits(msg.Round).BitArrayByBlockID(msg.BlockID)
default:
conR.Logger.Error("Bad VoteSetBitsMessage field Type")
return
}
src.TrySend(VoteSetBitsChannel, struct{ ConsensusMessage }{&VoteSetBitsMessage{
Height: msg.Height,
Round: msg.Round,
Type: msg.Type,
BlockID: msg.BlockID,
Votes: ourVotes,
}})
case *ProposalHeartbeatMessage:
hb := msg.Heartbeat
conR.Logger.Debug("Received proposal heartbeat message",
"height", hb.Height, "round", hb.Round, "sequence", hb.Sequence,
"valIdx", hb.ValidatorIndex, "valAddr", hb.ValidatorAddress)
default:
conR.Logger.Error(cmn.Fmt("Unknown message type %v", reflect.TypeOf(msg)))
}
case DataChannel:
if conR.FastSync() {
conR.Logger.Info("Ignoring message received during fastSync", "msg", msg)
return
}
switch msg := msg.(type) {
case *ProposalMessage:
ps.SetHasProposal(msg.Proposal)
conR.conS.peerMsgQueue <- msgInfo{msg, src.Key()}
case *ProposalPOLMessage:
ps.ApplyProposalPOLMessage(msg)
case *BlockPartMessage:
ps.SetHasProposalBlockPart(msg.Height, msg.Round, msg.Part.Index)
conR.conS.peerMsgQueue <- msgInfo{msg, src.Key()}
default:
conR.Logger.Error(cmn.Fmt("Unknown message type %v", reflect.TypeOf(msg)))
}
case VoteChannel:
if conR.FastSync() {
conR.Logger.Info("Ignoring message received during fastSync", "msg", msg)
return
}
switch msg := msg.(type) {
case *VoteMessage:
cs := conR.conS
cs.mtx.Lock()
height, valSize, lastCommitSize := cs.Height, cs.Validators.Size(), cs.LastCommit.Size()
cs.mtx.Unlock()
ps.EnsureVoteBitArrays(height, valSize)
ps.EnsureVoteBitArrays(height-1, lastCommitSize)
ps.SetHasVote(msg.Vote)
cs.peerMsgQueue <- msgInfo{msg, src.Key()}
default:
// don't punish (leave room for soft upgrades)
conR.Logger.Error(cmn.Fmt("Unknown message type %v", reflect.TypeOf(msg)))
}
case VoteSetBitsChannel:
if conR.FastSync() {
conR.Logger.Info("Ignoring message received during fastSync", "msg", msg)
return
}
switch msg := msg.(type) {
case *VoteSetBitsMessage:
cs := conR.conS
cs.mtx.Lock()
height, votes := cs.Height, cs.Votes
cs.mtx.Unlock()
if height == msg.Height {
var ourVotes *cmn.BitArray
switch msg.Type {
case types.VoteTypePrevote:
ourVotes = votes.Prevotes(msg.Round).BitArrayByBlockID(msg.BlockID)
case types.VoteTypePrecommit:
ourVotes = votes.Precommits(msg.Round).BitArrayByBlockID(msg.BlockID)
default:
conR.Logger.Error("Bad VoteSetBitsMessage field Type")
return
}
ps.ApplyVoteSetBitsMessage(msg, ourVotes)
} else {
ps.ApplyVoteSetBitsMessage(msg, nil)
}
default:
// don't punish (leave room for soft upgrades)
conR.Logger.Error(cmn.Fmt("Unknown message type %v", reflect.TypeOf(msg)))
}
default:
conR.Logger.Error(cmn.Fmt("Unknown chId %X", chID))
}
if err != nil {
conR.Logger.Error("Error in Receive()", "err", err)
}
}
// SetEventBus sets event bus.
func (conR *ConsensusReactor) SetEventBus(b *types.EventBus) {
conR.eventBus = b
conR.conS.SetEventBus(b)
}
// FastSync returns whether the consensus reactor is in fast-sync mode.
func (conR *ConsensusReactor) FastSync() bool {
conR.mtx.RLock()
defer conR.mtx.RUnlock()
return conR.fastSync
}
//--------------------------------------
// broadcastNewRoundStepsAndVotes subscribes for new round steps and votes
// using the event bus and broadcasts events to peers upon receiving them.
func (conR *ConsensusReactor) broadcastNewRoundStepsAndVotes() error {
subscriber := "consensus-reactor"
ctx := context.Background()
// new round steps
stepsCh := make(chan interface{})
err := conR.eventBus.Subscribe(ctx, subscriber, types.EventQueryNewRoundStep, stepsCh)
if err != nil {
return errors.Wrapf(err, "failed to subscribe %s to %s", subscriber, types.EventQueryNewRoundStep)
}
// votes
votesCh := make(chan interface{})
err = conR.eventBus.Subscribe(ctx, subscriber, types.EventQueryVote, votesCh)
if err != nil {
return errors.Wrapf(err, "failed to subscribe %s to %s", subscriber, types.EventQueryVote)
}
go func() {
for {
select {
case data, ok := <-stepsCh:
if ok { // a receive from a closed channel returns the zero value immediately
edrs := data.(types.TMEventData).Unwrap().(types.EventDataRoundState)
conR.broadcastNewRoundStep(edrs.RoundState.(*cstypes.RoundState))
}
case data, ok := <-votesCh:
if ok {
edv := data.(types.TMEventData).Unwrap().(types.EventDataVote)
conR.broadcastHasVoteMessage(edv.Vote)
}
case <-conR.Quit:
conR.eventBus.UnsubscribeAll(ctx, subscriber)
return
}
}
}()
return nil
}
func (conR *ConsensusReactor) broadcastProposalHeartbeatMessage(heartbeat types.EventDataProposalHeartbeat) {
hb := heartbeat.Heartbeat
conR.Logger.Debug("Broadcasting proposal heartbeat message",
"height", hb.Height, "round", hb.Round, "sequence", hb.Sequence)
msg := &ProposalHeartbeatMessage{hb}
conR.Switch.Broadcast(StateChannel, struct{ ConsensusMessage }{msg})
}
func (conR *ConsensusReactor) broadcastNewRoundStep(rs *cstypes.RoundState) {
nrsMsg, csMsg := makeRoundStepMessages(rs)
if nrsMsg != nil {
conR.Switch.Broadcast(StateChannel, struct{ ConsensusMessage }{nrsMsg})
}
if csMsg != nil {
conR.Switch.Broadcast(StateChannel, struct{ ConsensusMessage }{csMsg})
}
}
// Broadcasts HasVoteMessage to peers that care.
func (conR *ConsensusReactor) broadcastHasVoteMessage(vote *types.Vote) {
msg := &HasVoteMessage{
Height: vote.Height,
Round: vote.Round,
Type: vote.Type,
Index: vote.ValidatorIndex,
}
conR.Switch.Broadcast(StateChannel, struct{ ConsensusMessage }{msg})
/*
// TODO: Make this broadcast more selective.
for _, peer := range conR.Switch.Peers().List() {
ps := peer.Get(PeerStateKey).(*PeerState)
prs := ps.GetRoundState()
if prs.Height == vote.Height {
// TODO: Also filter on round?
peer.TrySend(StateChannel, struct{ ConsensusMessage }{msg})
} else {
// Height doesn't match
// TODO: check a field, maybe CatchupCommitRound?
// TODO: But that requires changing the struct field comment.
}
}
*/
}
func makeRoundStepMessages(rs *cstypes.RoundState) (nrsMsg *NewRoundStepMessage, csMsg *CommitStepMessage) {
nrsMsg = &NewRoundStepMessage{
Height: rs.Height,
Round: rs.Round,
Step: rs.Step,
SecondsSinceStartTime: int(time.Since(rs.StartTime).Seconds()),
LastCommitRound: rs.LastCommit.Round(),
}
if rs.Step == cstypes.RoundStepCommit {
csMsg = &CommitStepMessage{
Height: rs.Height,
BlockPartsHeader: rs.ProposalBlockParts.Header(),
BlockParts: rs.ProposalBlockParts.BitArray(),
}
}
return
}
func (conR *ConsensusReactor) sendNewRoundStepMessages(peer p2p.Peer) {
rs := conR.conS.GetRoundState()
nrsMsg, csMsg := makeRoundStepMessages(rs)
if nrsMsg != nil {
peer.Send(StateChannel, struct{ ConsensusMessage }{nrsMsg})
}
if csMsg != nil {
peer.Send(StateChannel, struct{ ConsensusMessage }{csMsg})
}
}
func (conR *ConsensusReactor) gossipDataRoutine(peer p2p.Peer, ps *PeerState) {
logger := conR.Logger.With("peer", peer)
OUTER_LOOP:
for {
// Manage disconnects from self or peer.
if !peer.IsRunning() || !conR.IsRunning() {
logger.Info("Stopping gossipDataRoutine for peer")
return
}
rs := conR.conS.GetRoundState()
prs := ps.GetRoundState()
// Send proposal Block parts?
if rs.ProposalBlockParts.HasHeader(prs.ProposalBlockPartsHeader) {
if index, ok := rs.ProposalBlockParts.BitArray().Sub(prs.ProposalBlockParts.Copy()).PickRandom(); ok {
part := rs.ProposalBlockParts.GetPart(index)
msg := &BlockPartMessage{
Height: rs.Height, // This tells peer that this part applies to us.
Round: rs.Round, // This tells peer that this part applies to us.
Part: part,
}
logger.Debug("Sending block part", "height", prs.Height, "round", prs.Round)
if peer.Send(DataChannel, struct{ ConsensusMessage }{msg}) {
ps.SetHasProposalBlockPart(prs.Height, prs.Round, index)
}
continue OUTER_LOOP
}
}
// If the peer is on a previous height, help catch up.
if (0 < prs.Height) && (prs.Height < rs.Height) {
heightLogger := logger.With("height", prs.Height)
conR.gossipDataForCatchup(heightLogger, rs, prs, ps, peer)
continue OUTER_LOOP
}
// If height and round don't match, sleep.
if (rs.Height != prs.Height) || (rs.Round != prs.Round) {
//logger.Info("Peer Height|Round mismatch, sleeping", "peerHeight", prs.Height, "peerRound", prs.Round, "peer", peer)
time.Sleep(conR.conS.config.PeerGossipSleep())
continue OUTER_LOOP
}
// By here, height and round match.
// Proposal block parts were already matched and sent if any were wanted.
// (These can match on hash so the round doesn't matter)
// Now consider sending other things, like the Proposal itself.
// Send Proposal && ProposalPOL BitArray?
if rs.Proposal != nil && !prs.Proposal {
// Proposal: share the proposal metadata with peer.
{
msg := &ProposalMessage{Proposal: rs.Proposal}
logger.Debug("Sending proposal", "height", prs.Height, "round", prs.Round)
if peer.Send(DataChannel, struct{ ConsensusMessage }{msg}) {
ps.SetHasProposal(rs.Proposal)
}
}
// ProposalPOL: lets peer know which POL votes we have so far.
// Peer must receive ProposalMessage first.
// rs.Proposal was validated, so rs.Proposal.POLRound <= rs.Round,
// so we definitely have rs.Votes.Prevotes(rs.Proposal.POLRound).
if 0 <= rs.Proposal.POLRound {
msg := &ProposalPOLMessage{
Height: rs.Height,
ProposalPOLRound: rs.Proposal.POLRound,
ProposalPOL: rs.Votes.Prevotes(rs.Proposal.POLRound).BitArray(),
}
logger.Debug("Sending POL", "height", prs.Height, "round", prs.Round)
peer.Send(DataChannel, struct{ ConsensusMessage }{msg})
}
continue OUTER_LOOP
}
// Nothing to do. Sleep.
time.Sleep(conR.conS.config.PeerGossipSleep())
continue OUTER_LOOP
}
}
func (conR *ConsensusReactor) gossipDataForCatchup(logger log.Logger, rs *cstypes.RoundState,
prs *cstypes.PeerRoundState, ps *PeerState, peer p2p.Peer) {
if index, ok := prs.ProposalBlockParts.Not().PickRandom(); ok {
// Ensure that the peer's PartSetHeader is correct
blockMeta := conR.conS.blockStore.LoadBlockMeta(prs.Height)
if blockMeta == nil {
logger.Error("Failed to load block meta",
"ourHeight", rs.Height, "blockstoreHeight", conR.conS.blockStore.Height())
time.Sleep(conR.conS.config.PeerGossipSleep())
return
} else if !blockMeta.BlockID.PartsHeader.Equals(prs.ProposalBlockPartsHeader) {
logger.Info("Peer ProposalBlockPartsHeader mismatch, sleeping",
"blockPartsHeader", blockMeta.BlockID.PartsHeader, "peerBlockPartsHeader", prs.ProposalBlockPartsHeader)
time.Sleep(conR.conS.config.PeerGossipSleep())
return
}
// Load the part
part := conR.conS.blockStore.LoadBlockPart(prs.Height, index)
if part == nil {
logger.Error("Could not load part", "index", index,
"blockPartsHeader", blockMeta.BlockID.PartsHeader, "peerBlockPartsHeader", prs.ProposalBlockPartsHeader)
time.Sleep(conR.conS.config.PeerGossipSleep())
return
}
// Send the part
msg := &BlockPartMessage{
Height: prs.Height, // Not our height, so it doesn't matter.
Round: prs.Round, // Not our height, so it doesn't matter.
Part: part,
}
logger.Debug("Sending block part for catchup", "round", prs.Round)
if peer.Send(DataChannel, struct{ ConsensusMessage }{msg}) {
ps.SetHasProposalBlockPart(prs.Height, prs.Round, index)
}
return
} else {
//logger.Info("No parts to send in catch-up, sleeping")
time.Sleep(conR.conS.config.PeerGossipSleep())
return
}
}
func (conR *ConsensusReactor) gossipVotesRoutine(peer p2p.Peer, ps *PeerState) {
logger := conR.Logger.With("peer", peer)
// Simple hack to throttle logs upon sleep.
var sleeping = 0
OUTER_LOOP:
for {
// Manage disconnects from self or peer.
if !peer.IsRunning() || !conR.IsRunning() {
logger.Info("Stopping gossipVotesRoutine for peer")
return
}
rs := conR.conS.GetRoundState()
prs := ps.GetRoundState()
switch sleeping {
case 1: // First sleep
sleeping = 2
case 2: // No more sleep
sleeping = 0
}
//logger.Debug("gossipVotesRoutine", "rsHeight", rs.Height, "rsRound", rs.Round,
// "prsHeight", prs.Height, "prsRound", prs.Round, "prsStep", prs.Step)
// If height matches, then send LastCommit, Prevotes, Precommits.
if rs.Height == prs.Height {
heightLogger := logger.With("height", prs.Height)
if conR.gossipVotesForHeight(heightLogger, rs, prs, ps) {
continue OUTER_LOOP
}
}
// Special catchup logic.
// If peer is lagging by height 1, send LastCommit.
if prs.Height != 0 && rs.Height == prs.Height+1 {
if ps.PickSendVote(rs.LastCommit) {
logger.Debug("Picked rs.LastCommit to send", "height", prs.Height)
continue OUTER_LOOP
}
}
// Catchup logic
// If peer is lagging by more than 1, send Commit.
if prs.Height != 0 && rs.Height >= prs.Height+2 {
// Load the block commit for prs.Height,
// which contains precommit signatures for prs.Height.
commit := conR.conS.blockStore.LoadBlockCommit(prs.Height)
logger.Info("Loaded BlockCommit for catch-up", "height", prs.Height, "commit", commit)
if ps.PickSendVote(commit) {
logger.Debug("Picked Catchup commit to send", "height", prs.Height)
continue OUTER_LOOP
}
}
if sleeping == 0 {
// We sent nothing. Sleep...
sleeping = 1
logger.Debug("No votes to send, sleeping", "rs.Height", rs.Height, "prs.Height", prs.Height,
"localPV", rs.Votes.Prevotes(rs.Round).BitArray(), "peerPV", prs.Prevotes,
"localPC", rs.Votes.Precommits(rs.Round).BitArray(), "peerPC", prs.Precommits)
} else if sleeping == 2 {
// Continued sleep...
sleeping = 1
}
time.Sleep(conR.conS.config.PeerGossipSleep())
continue OUTER_LOOP
}
}
func (conR *ConsensusReactor) gossipVotesForHeight(logger log.Logger, rs *cstypes.RoundState, prs *cstypes.PeerRoundState, ps *PeerState) bool {
// If there are lastCommits to send...
if prs.Step == cstypes.RoundStepNewHeight {
if ps.PickSendVote(rs.LastCommit) {
logger.Debug("Picked rs.LastCommit to send")
return true
}
}
// If there are prevotes to send...
if prs.Step <= cstypes.RoundStepPrevote && prs.Round != -1 && prs.Round <= rs.Round {
if ps.PickSendVote(rs.Votes.Prevotes(prs.Round)) {
logger.Debug("Picked rs.Prevotes(prs.Round) to send", "round", prs.Round)
return true
}
}
// If there are precommits to send...
if prs.Step <= cstypes.RoundStepPrecommit && prs.Round != -1 && prs.Round <= rs.Round {
if ps.PickSendVote(rs.Votes.Precommits(prs.Round)) {
logger.Debug("Picked rs.Precommits(prs.Round) to send", "round", prs.Round)
return true
}
}
// If there are POLPrevotes to send...
if prs.ProposalPOLRound != -1 {
if polPrevotes := rs.Votes.Prevotes(prs.ProposalPOLRound); polPrevotes != nil {
if ps.PickSendVote(polPrevotes) {
logger.Debug("Picked rs.Prevotes(prs.ProposalPOLRound) to send",
"round", prs.ProposalPOLRound)
return true
}
}
}
return false
}
// NOTE: `queryMaj23Routine` has a simple crude design since it only comes
// into play for liveness when there's a signature DDoS attack happening.
func (conR *ConsensusReactor) queryMaj23Routine(peer p2p.Peer, ps *PeerState) {
logger := conR.Logger.With("peer", peer)
OUTER_LOOP:
for {
// Manage disconnects from self or peer.
if !peer.IsRunning() || !conR.IsRunning() {
logger.Info("Stopping queryMaj23Routine for peer")
return
}
// Maybe send Height/Round/Prevotes
{
rs := conR.conS.GetRoundState()
prs := ps.GetRoundState()
if rs.Height == prs.Height {
if maj23, ok := rs.Votes.Prevotes(prs.Round).TwoThirdsMajority(); ok {
peer.TrySend(StateChannel, struct{ ConsensusMessage }{&VoteSetMaj23Message{
Height: prs.Height,
Round: prs.Round,
Type: types.VoteTypePrevote,
BlockID: maj23,
}})
time.Sleep(conR.conS.config.PeerQueryMaj23Sleep())
}
}
}
// Maybe send Height/Round/Precommits
{
rs := conR.conS.GetRoundState()
prs := ps.GetRoundState()
if rs.Height == prs.Height {
if maj23, ok := rs.Votes.Precommits(prs.Round).TwoThirdsMajority(); ok {
peer.TrySend(StateChannel, struct{ ConsensusMessage }{&VoteSetMaj23Message{
Height: prs.Height,
Round: prs.Round,
Type: types.VoteTypePrecommit,
BlockID: maj23,
}})
time.Sleep(conR.conS.config.PeerQueryMaj23Sleep())
}
}
}
// Maybe send Height/Round/ProposalPOL
{
rs := conR.conS.GetRoundState()
prs := ps.GetRoundState()
if rs.Height == prs.Height && prs.ProposalPOLRound >= 0 {
if maj23, ok := rs.Votes.Prevotes(prs.ProposalPOLRound).TwoThirdsMajority(); ok {
peer.TrySend(StateChannel, struct{ ConsensusMessage }{&VoteSetMaj23Message{
Height: prs.Height,
Round: prs.ProposalPOLRound,
Type: types.VoteTypePrevote,
BlockID: maj23,
}})
time.Sleep(conR.conS.config.PeerQueryMaj23Sleep())
}
}
}
// Little point sending LastCommitRound/LastCommit,
// These are fleeting and non-blocking.
// Maybe send Height/CatchupCommitRound/CatchupCommit.
{
prs := ps.GetRoundState()
if prs.CatchupCommitRound != -1 && 0 < prs.Height && prs.Height <= conR.conS.blockStore.Height() {
commit := conR.conS.LoadCommit(prs.Height)
peer.TrySend(StateChannel, struct{ ConsensusMessage }{&VoteSetMaj23Message{
Height: prs.Height,
Round: commit.Round(),
Type: types.VoteTypePrecommit,
BlockID: commit.BlockID,
}})
time.Sleep(conR.conS.config.PeerQueryMaj23Sleep())
}
}
time.Sleep(conR.conS.config.PeerQueryMaj23Sleep())
continue OUTER_LOOP
}
}
// String returns a string representation of the ConsensusReactor.
// NOTE: For now, it is just a hard-coded string to avoid accessing unprotected shared variables.
// TODO: improve!
func (conR *ConsensusReactor) String() string {
// better not to access shared variables
return "ConsensusReactor" // conR.StringIndented("")
}
// StringIndented returns an indented string representation of the ConsensusReactor
func (conR *ConsensusReactor) StringIndented(indent string) string {
s := "ConsensusReactor{\n"
s += indent + " " + conR.conS.StringIndented(indent+" ") + "\n"
for _, peer := range conR.Switch.Peers().List() {
ps := peer.Get(types.PeerStateKey).(*PeerState)
s += indent + " " + ps.StringIndented(indent+" ") + "\n"
}
s += indent + "}"
return s
}
//-----------------------------------------------------------------------------
var (
ErrPeerStateHeightRegression = errors.New("Error peer state height regression")
ErrPeerStateInvalidStartTime = errors.New("Error peer state invalid startTime")
)
// PeerState contains the known state of a peer, including its connection
// and threadsafe access to its PeerRoundState.
type PeerState struct {
Peer p2p.Peer
logger log.Logger
mtx sync.Mutex
cstypes.PeerRoundState
}
// NewPeerState returns a new PeerState for the given Peer
func NewPeerState(peer p2p.Peer) *PeerState {
return &PeerState{
Peer: peer,
logger: log.NewNopLogger(),
PeerRoundState: cstypes.PeerRoundState{
Round: -1,
ProposalPOLRound: -1,
LastCommitRound: -1,
CatchupCommitRound: -1,
},
}
}
func (ps *PeerState) SetLogger(logger log.Logger) *PeerState {
ps.logger = logger
return ps
}
// GetRoundState returns an atomic snapshot of the PeerRoundState.
// There's no point in mutating it since it won't change PeerState.
func (ps *PeerState) GetRoundState() *cstypes.PeerRoundState {
ps.mtx.Lock()
defer ps.mtx.Unlock()
prs := ps.PeerRoundState // copy
return &prs
}
// GetHeight returns an atomic snapshot of the PeerRoundState's height
// used by the mempool to ensure peers are caught up before broadcasting new txs
func (ps *PeerState) GetHeight() int {
ps.mtx.Lock()
defer ps.mtx.Unlock()
return ps.PeerRoundState.Height
}
// SetHasProposal sets the given proposal as known for the peer.
func (ps *PeerState) SetHasProposal(proposal *types.Proposal) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if ps.Height != proposal.Height || ps.Round != proposal.Round {
return
}
if ps.Proposal {
return
}
ps.Proposal = true
ps.ProposalBlockPartsHeader = proposal.BlockPartsHeader
ps.ProposalBlockParts = cmn.NewBitArray(proposal.BlockPartsHeader.Total)
ps.ProposalPOLRound = proposal.POLRound
ps.ProposalPOL = nil // Nil until ProposalPOLMessage received.
}
// SetHasProposalBlockPart sets the given block part index as known for the peer.
func (ps *PeerState) SetHasProposalBlockPart(height int, round int, index int) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if ps.Height != height || ps.Round != round {
return
}
ps.ProposalBlockParts.SetIndex(index, true)
}
// PickSendVote picks a vote and sends it to the peer.
// Returns true if vote was sent.
func (ps *PeerState) PickSendVote(votes types.VoteSetReader) bool {
if vote, ok := ps.PickVoteToSend(votes); ok {
msg := &VoteMessage{vote}
return ps.Peer.Send(VoteChannel, struct{ ConsensusMessage }{msg})
}
return false
}
// PickVoteToSend picks a vote to send to the peer.
// Returns true if a vote was picked.
// NOTE: `votes` must be the correct Size() for the Height().
func (ps *PeerState) PickVoteToSend(votes types.VoteSetReader) (vote *types.Vote, ok bool) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if votes.Size() == 0 {
return nil, false
}
height, round, type_, size := votes.Height(), votes.Round(), votes.Type(), votes.Size()
// Lazily set data using 'votes'.
if votes.IsCommit() {
ps.ensureCatchupCommitRound(height, round, size)
}
ps.ensureVoteBitArrays(height, size)
psVotes := ps.getVoteBitArray(height, round, type_)
if psVotes == nil {
return nil, false // Not something worth sending
}
if index, ok := votes.BitArray().Sub(psVotes).PickRandom(); ok {
ps.setHasVote(height, round, type_, index)
return votes.GetByIndex(index), true
}
return nil, false
}
func (ps *PeerState) getVoteBitArray(height, round int, type_ byte) *cmn.BitArray {
if !types.IsVoteTypeValid(type_) {
return nil
}
if ps.Height == height {
if ps.Round == round {
switch type_ {
case types.VoteTypePrevote:
return ps.Prevotes
case types.VoteTypePrecommit:
return ps.Precommits
}
}
if ps.CatchupCommitRound == round {
switch type_ {
case types.VoteTypePrevote:
return nil
case types.VoteTypePrecommit:
return ps.CatchupCommit
}
}
if ps.ProposalPOLRound == round {
switch type_ {
case types.VoteTypePrevote:
return ps.ProposalPOL
case types.VoteTypePrecommit:
return nil
}
}
return nil
}
if ps.Height == height+1 {
if ps.LastCommitRound == round {
switch type_ {
case types.VoteTypePrevote:
return nil
case types.VoteTypePrecommit:
return ps.LastCommit
}
}
return nil
}
return nil
}
// 'round': A round for which we have a +2/3 commit.
func (ps *PeerState) ensureCatchupCommitRound(height, round int, numValidators int) {
if ps.Height != height {
return
}
/*
NOTE: This is wrong, 'round' could change.
e.g. if orig round is not the same as block LastCommit round.
if ps.CatchupCommitRound != -1 && ps.CatchupCommitRound != round {
cmn.PanicSanity(cmn.Fmt("Conflicting CatchupCommitRound. Height: %v, Orig: %v, New: %v", height, ps.CatchupCommitRound, round))
}
*/
if ps.CatchupCommitRound == round {
return // Nothing to do!
}
ps.CatchupCommitRound = round
if round == ps.Round {
ps.CatchupCommit = ps.Precommits
} else {
ps.CatchupCommit = cmn.NewBitArray(numValidators)
}
}
// EnsureVoteVitArrays ensures the bit-arrays have been allocated for tracking
// what votes this peer has received.
// NOTE: It's important to make sure that numValidators actually matches
// what the node sees as the number of validators for height.
func (ps *PeerState) EnsureVoteBitArrays(height int, numValidators int) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
ps.ensureVoteBitArrays(height, numValidators)
}
func (ps *PeerState) ensureVoteBitArrays(height int, numValidators int) {
if ps.Height == height {
if ps.Prevotes == nil {
ps.Prevotes = cmn.NewBitArray(numValidators)
}
if ps.Precommits == nil {
ps.Precommits = cmn.NewBitArray(numValidators)
}
if ps.CatchupCommit == nil {
ps.CatchupCommit = cmn.NewBitArray(numValidators)
}
if ps.ProposalPOL == nil {
ps.ProposalPOL = cmn.NewBitArray(numValidators)
}
} else if ps.Height == height+1 {
if ps.LastCommit == nil {
ps.LastCommit = cmn.NewBitArray(numValidators)
}
}
}
// SetHasVote sets the given vote as known by the peer
func (ps *PeerState) SetHasVote(vote *types.Vote) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
ps.setHasVote(vote.Height, vote.Round, vote.Type, vote.ValidatorIndex)
}
func (ps *PeerState) setHasVote(height int, round int, type_ byte, index int) {
logger := ps.logger.With("peerRound", ps.Round, "height", height, "round", round)
logger.Debug("setHasVote(LastCommit)", "lastCommit", ps.LastCommit, "index", index)
// NOTE: some may be nil BitArrays -> no side effects.
psVotes := ps.getVoteBitArray(height, round, type_)
if psVotes != nil {
psVotes.SetIndex(index, true)
}
}
// ApplyNewRoundStepMessage updates the peer state for the new round.
func (ps *PeerState) ApplyNewRoundStepMessage(msg *NewRoundStepMessage) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
// Ignore duplicates or decreases
if CompareHRS(msg.Height, msg.Round, msg.Step, ps.Height, ps.Round, ps.Step) <= 0 {
return
}
// Just remember these values.
psHeight := ps.Height
psRound := ps.Round
//psStep := ps.Step
psCatchupCommitRound := ps.CatchupCommitRound
psCatchupCommit := ps.CatchupCommit
startTime := time.Now().Add(-1 * time.Duration(msg.SecondsSinceStartTime) * time.Second)
ps.Height = msg.Height
ps.Round = msg.Round
ps.Step = msg.Step
ps.StartTime = startTime
if psHeight != msg.Height || psRound != msg.Round {
ps.Proposal = false
ps.ProposalBlockPartsHeader = types.PartSetHeader{}
ps.ProposalBlockParts = nil
ps.ProposalPOLRound = -1
ps.ProposalPOL = nil
// We'll update the BitArray capacity later.
ps.Prevotes = nil
ps.Precommits = nil
}
if psHeight == msg.Height && psRound != msg.Round && msg.Round == psCatchupCommitRound {
// Peer caught up to CatchupCommitRound.
// Preserve psCatchupCommit!
// NOTE: We prefer to use prs.Precommits if
// pr.Round matches pr.CatchupCommitRound.
ps.Precommits = psCatchupCommit
}
if psHeight != msg.Height {
// Shift Precommits to LastCommit.
if psHeight+1 == msg.Height && psRound == msg.LastCommitRound {
ps.LastCommitRound = msg.LastCommitRound
ps.LastCommit = ps.Precommits
} else {
ps.LastCommitRound = msg.LastCommitRound
ps.LastCommit = nil
}
// We'll update the BitArray capacity later.
ps.CatchupCommitRound = -1
ps.CatchupCommit = nil
}
}
// ApplyCommitStepMessage updates the peer state for the new commit.
func (ps *PeerState) ApplyCommitStepMessage(msg *CommitStepMessage) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if ps.Height != msg.Height {
return
}
ps.ProposalBlockPartsHeader = msg.BlockPartsHeader
ps.ProposalBlockParts = msg.BlockParts
}
// ApplyProposalPOLMessage updates the peer state for the new proposal POL.
func (ps *PeerState) ApplyProposalPOLMessage(msg *ProposalPOLMessage) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if ps.Height != msg.Height {
return
}
if ps.ProposalPOLRound != msg.ProposalPOLRound {
return
}
// TODO: Merge onto existing ps.ProposalPOL?
// We might have sent some prevotes in the meantime.
ps.ProposalPOL = msg.ProposalPOL
}
// ApplyHasVoteMessage updates the peer state for the new vote.
func (ps *PeerState) ApplyHasVoteMessage(msg *HasVoteMessage) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
if ps.Height != msg.Height {
return
}
ps.setHasVote(msg.Height, msg.Round, msg.Type, msg.Index)
}
// ApplyVoteSetBitsMessage updates the peer state for the bit-array of votes
// it claims to have for the corresponding BlockID.
// `ourVotes` is a BitArray of votes we have for msg.BlockID
// NOTE: if ourVotes is nil (e.g. msg.Height < rs.Height),
// we conservatively overwrite ps's votes w/ msg.Votes.
func (ps *PeerState) ApplyVoteSetBitsMessage(msg *VoteSetBitsMessage, ourVotes *cmn.BitArray) {
ps.mtx.Lock()
defer ps.mtx.Unlock()
votes := ps.getVoteBitArray(msg.Height, msg.Round, msg.Type)
if votes != nil {
if ourVotes == nil {
votes.Update(msg.Votes)
} else {
otherVotes := votes.Sub(ourVotes)
hasVotes := otherVotes.Or(msg.Votes)
votes.Update(hasVotes)
}
}
}
// String returns a string representation of the PeerState
func (ps *PeerState) String() string {
return ps.StringIndented("")
}
// StringIndented returns a string representation of the PeerState
func (ps *PeerState) StringIndented(indent string) string {
return fmt.Sprintf(`PeerState{
%s Key %v
%s PRS %v
%s}`,
indent, ps.Peer.Key(),
indent, ps.PeerRoundState.StringIndented(indent+" "),
indent)
}
//-----------------------------------------------------------------------------
// Messages
const (
msgTypeNewRoundStep = byte(0x01)
msgTypeCommitStep = byte(0x02)
msgTypeProposal = byte(0x11)
msgTypeProposalPOL = byte(0x12)
msgTypeBlockPart = byte(0x13) // both block & POL
msgTypeVote = byte(0x14)
msgTypeHasVote = byte(0x15)
msgTypeVoteSetMaj23 = byte(0x16)
msgTypeVoteSetBits = byte(0x17)
msgTypeProposalHeartbeat = byte(0x20)
)
// ConsensusMessage is a message that can be sent and received on the ConsensusReactor
type ConsensusMessage interface{}
var _ = wire.RegisterInterface(
struct{ ConsensusMessage }{},
wire.ConcreteType{&NewRoundStepMessage{}, msgTypeNewRoundStep},
wire.ConcreteType{&CommitStepMessage{}, msgTypeCommitStep},
wire.ConcreteType{&ProposalMessage{}, msgTypeProposal},
wire.ConcreteType{&ProposalPOLMessage{}, msgTypeProposalPOL},
wire.ConcreteType{&BlockPartMessage{}, msgTypeBlockPart},
wire.ConcreteType{&VoteMessage{}, msgTypeVote},
wire.ConcreteType{&HasVoteMessage{}, msgTypeHasVote},
wire.ConcreteType{&VoteSetMaj23Message{}, msgTypeVoteSetMaj23},
wire.ConcreteType{&VoteSetBitsMessage{}, msgTypeVoteSetBits},
wire.ConcreteType{&ProposalHeartbeatMessage{}, msgTypeProposalHeartbeat},
)
// DecodeMessage decodes the given bytes into a ConsensusMessage.
// TODO: check for unnecessary extra bytes at the end.
func DecodeMessage(bz []byte) (msgType byte, msg ConsensusMessage, err error) {
msgType = bz[0]
n := new(int)
r := bytes.NewReader(bz)
msgI := wire.ReadBinary(struct{ ConsensusMessage }{}, r, maxConsensusMessageSize, n, &err)
msg = msgI.(struct{ ConsensusMessage }).ConsensusMessage
return
}
//-------------------------------------
// NewRoundStepMessage is sent for every step taken in the ConsensusState.
// For every height/round/step transition
type NewRoundStepMessage struct {
Height int
Round int
Step cstypes.RoundStepType
SecondsSinceStartTime int
LastCommitRound int
}
// String returns a string representation.
func (m *NewRoundStepMessage) String() string {
return fmt.Sprintf("[NewRoundStep H:%v R:%v S:%v LCR:%v]",
m.Height, m.Round, m.Step, m.LastCommitRound)
}
//-------------------------------------
// CommitStepMessage is sent when a block is committed.
type CommitStepMessage struct {
Height int
BlockPartsHeader types.PartSetHeader
BlockParts *cmn.BitArray
}
// String returns a string representation.
func (m *CommitStepMessage) String() string {
return fmt.Sprintf("[CommitStep H:%v BP:%v BA:%v]", m.Height, m.BlockPartsHeader, m.BlockParts)
}
//-------------------------------------
// ProposalMessage is sent when a new block is proposed.
type ProposalMessage struct {
Proposal *types.Proposal
}
// String returns a string representation.
func (m *ProposalMessage) String() string {
return fmt.Sprintf("[Proposal %v]", m.Proposal)
}
//-------------------------------------
// ProposalPOLMessage is sent when a previous proposal is re-proposed.
type ProposalPOLMessage struct {
Height int
ProposalPOLRound int
ProposalPOL *cmn.BitArray
}
// String returns a string representation.
func (m *ProposalPOLMessage) String() string {
return fmt.Sprintf("[ProposalPOL H:%v POLR:%v POL:%v]", m.Height, m.ProposalPOLRound, m.ProposalPOL)
}
//-------------------------------------
// BlockPartMessage is sent when gossipping a piece of the proposed block.
type BlockPartMessage struct {
Height int
Round int
Part *types.Part
}
// String returns a string representation.
func (m *BlockPartMessage) String() string {
return fmt.Sprintf("[BlockPart H:%v R:%v P:%v]", m.Height, m.Round, m.Part)
}
//-------------------------------------
// VoteMessage is sent when voting for a proposal (or lack thereof).
type VoteMessage struct {
Vote *types.Vote
}
// String returns a string representation.
func (m *VoteMessage) String() string {
return fmt.Sprintf("[Vote %v]", m.Vote)
}
//-------------------------------------
// HasVoteMessage is sent to indicate that a particular vote has been received.
type HasVoteMessage struct {
Height int
Round int
Type byte
Index int
}
// String returns a string representation.
func (m *HasVoteMessage) String() string {
return fmt.Sprintf("[HasVote VI:%v V:{%v/%02d/%v} VI:%v]", m.Index, m.Height, m.Round, m.Type, m.Index)
}
//-------------------------------------
// VoteSetMaj23Message is sent to indicate that a given BlockID has seen +2/3 votes.
type VoteSetMaj23Message struct {
Height int
Round int
Type byte
BlockID types.BlockID
}
// String returns a string representation.
func (m *VoteSetMaj23Message) String() string {
return fmt.Sprintf("[VSM23 %v/%02d/%v %v]", m.Height, m.Round, m.Type, m.BlockID)
}
//-------------------------------------
// VoteSetBitsMessage is sent to communicate the bit-array of votes seen for the BlockID.
type VoteSetBitsMessage struct {
Height int
Round int
Type byte
BlockID types.BlockID
Votes *cmn.BitArray
}
// String returns a string representation.
func (m *VoteSetBitsMessage) String() string {
return fmt.Sprintf("[VSB %v/%02d/%v %v %v]", m.Height, m.Round, m.Type, m.BlockID, m.Votes)
}
//-------------------------------------
// ProposalHeartbeatMessage is sent to signal that a node is alive and waiting for transactions for a proposal.
type ProposalHeartbeatMessage struct {
Heartbeat *types.Heartbeat
}
// String returns a string representation.
func (m *ProposalHeartbeatMessage) String() string {
return fmt.Sprintf("[HEARTBEAT %v]", m.Heartbeat)
}