fluencelabs/tendermint
1
0
Fork 0
mirror of https://github.com/fluencelabs/tendermint synced 2025-06-12 12:51:22 +00:00
Code Issues Projects Releases Wiki Activity

new pubsub package

Browse Source 
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
This commit is contained in:
Anton Kaliaev
2017-06-26 19:00:30 +04:00
parent 7c85f15a6c
commit f6539737de
43 changed files with 1368 additions and 870 deletions
Download Patch File Download Diff File Expand all files Collapse all files

101
consensus/state.go
View File

@ -91,13 +91,13 @@ type ConsensusState struct {
internalMsgQueue chan msgInfo
timeoutTicker TimeoutTicker
// we use PubSub to trigger msg broadcasts in the reactor,
// we use eventBus to trigger msg broadcasts in the reactor,
// and to notify external subscribers, eg. through a websocket
evsw types.EventSwitch
eventBus *types.EventBus
// a Write-Ahead Log ensures we can recover from any kind of crash
// and helps us avoid signing conflicting votes
wal *WAL
wal WAL
replayMode bool // so we don't log signing errors during replay
doWALCatchup bool // determines if we even try to do the catchup
@ -125,6 +125,7 @@ func NewConsensusState(config *cfg.ConsensusConfig, state *sm.State, proxyAppCon
timeoutTicker: NewTimeoutTicker(),
done: make(chan struct{}),
doWALCatchup: true,
wal: nilWAL{},
}
// set function defaults (may be overwritten before calling Start)
cs.decideProposal = cs.defaultDecideProposal
@ -148,9 +149,9 @@ func (cs *ConsensusState) SetLogger(l log.Logger) {
cs.timeoutTicker.SetLogger(l)
}
// SetEventSwitch implements events.Eventable
func (cs *ConsensusState) SetEventSwitch(evsw types.EventSwitch) {
cs.evsw = evsw
// SetEventBus sets event bus.
func (cs *ConsensusState) SetEventBus(b *types.EventBus) {
cs.eventBus = b
}
// String returns a string.
@ -212,11 +213,16 @@ func (cs *ConsensusState) LoadCommit(height int) *types.Commit {
// OnStart implements cmn.Service.
// It loads the latest state via the WAL, and starts the timeout and receive routines.
func (cs *ConsensusState) OnStart() error {
walFile := cs.config.WalFile()
if err := cs.OpenWAL(walFile); err != nil {
cs.Logger.Error("Error loading ConsensusState wal", "err", err.Error())
return err
// we may set the WAL in testing before calling Start,
// so only OpenWAL if its still the nilWAL
if _, ok := cs.wal.(nilWAL); ok {
walFile := cs.config.WalFile()
wal, err := cs.OpenWAL(walFile)
if err != nil {
cs.Logger.Error("Error loading ConsensusState wal", "err", err.Error())
return err
}
cs.wal = wal
}
// we need the timeoutRoutine for replay so
@ -260,7 +266,7 @@ func (cs *ConsensusState) OnStop() {
cs.timeoutTicker.Stop()
// Make BaseService.Wait() wait until cs.wal.Wait()
if cs.wal != nil && cs.IsRunning() {
if cs.IsRunning() {
cs.wal.Wait()
}
}
@ -273,25 +279,22 @@ func (cs *ConsensusState) Wait() {
}
// OpenWAL opens a file to log all consensus messages and timeouts for deterministic accountability
func (cs *ConsensusState) OpenWAL(walFile string) (err error) {
err = cmn.EnsureDir(filepath.Dir(walFile), 0700)
func (cs *ConsensusState) OpenWAL(walFile string) (WAL, error) {
err := cmn.EnsureDir(filepath.Dir(walFile), 0700)
if err != nil {
cs.Logger.Error("Error ensuring ConsensusState wal dir", "err", err.Error())
return err
return nil, err
}
cs.mtx.Lock()
defer cs.mtx.Unlock()
wal, err := NewWAL(walFile, cs.config.WalLight)
if err != nil {
return err
return nil, err
}
wal.SetLogger(cs.Logger.With("wal", walFile))
if _, err := wal.Start(); err != nil {
return err
return nil, err
}
cs.wal = wal
return nil
return wal, nil
}
//------------------------------------------------------------
@ -480,9 +483,9 @@ func (cs *ConsensusState) newStep() {
rs := cs.RoundStateEvent()
cs.wal.Save(rs)
cs.nSteps += 1
// newStep is called by updateToStep in NewConsensusState before the evsw is set!
if cs.evsw != nil {
types.FireEventNewRoundStep(cs.evsw, rs)
// newStep is called by updateToStep in NewConsensusState before the eventBus is set!
if cs.eventBus != nil {
cs.eventBus.PublishEventNewRoundStep(rs)
}
}
@ -536,9 +539,7 @@ func (cs *ConsensusState) receiveRoutine(maxSteps int) {
// priv_val tracks LastSig
// close wal now that we're done writing to it
if cs.wal != nil {
cs.wal.Stop()
}
cs.wal.Stop()
close(cs.done)
return
@ -607,13 +608,13 @@ func (cs *ConsensusState) handleTimeout(ti timeoutInfo, rs cstypes.RoundState) {
case cstypes.RoundStepNewRound:
cs.enterPropose(ti.Height, 0)
case cstypes.RoundStepPropose:
types.FireEventTimeoutPropose(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventTimeoutPropose(cs.RoundStateEvent())
cs.enterPrevote(ti.Height, ti.Round)
case cstypes.RoundStepPrevoteWait:
types.FireEventTimeoutWait(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventTimeoutWait(cs.RoundStateEvent())
cs.enterPrecommit(ti.Height, ti.Round)
case cstypes.RoundStepPrecommitWait:
types.FireEventTimeoutWait(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventTimeoutWait(cs.RoundStateEvent())
cs.enterNewRound(ti.Height, ti.Round+1)
default:
panic(cmn.Fmt("Invalid timeout step: %v", ti.Step))
@ -673,7 +674,7 @@ func (cs *ConsensusState) enterNewRound(height int, round int) {
}
cs.Votes.SetRound(round + 1) // also track next round (round+1) to allow round-skipping
types.FireEventNewRound(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventNewRound(cs.RoundStateEvent())
// Wait for txs to be available in the mempool
// before we enterPropose in round 0. If the last block changed the app hash,
@ -726,8 +727,7 @@ func (cs *ConsensusState) proposalHeartbeat(height, round int) {
ValidatorIndex: valIndex,
}
cs.privValidator.SignHeartbeat(chainID, heartbeat)
heartbeatEvent := types.EventDataProposalHeartbeat{heartbeat}
types.FireEventProposalHeartbeat(cs.evsw, heartbeatEvent)
cs.eventBus.PublishEventProposalHeartbeat(types.EventDataProposalHeartbeat{heartbeat})
counter += 1
time.Sleep(proposalHeartbeatIntervalSeconds * time.Second)
}
@ -885,7 +885,7 @@ func (cs *ConsensusState) enterPrevote(height int, round int) {
// fire event for how we got here
if cs.isProposalComplete() {
types.FireEventCompleteProposal(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventCompleteProposal(cs.RoundStateEvent())
} else {
// we received +2/3 prevotes for a future round
// TODO: catchup event?
@ -987,7 +987,7 @@ func (cs *ConsensusState) enterPrecommit(height int, round int) {
}
// At this point +2/3 prevoted for a particular block or nil
types.FireEventPolka(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventPolka(cs.RoundStateEvent())
// the latest POLRound should be this round
polRound, _ := cs.Votes.POLInfo()
@ -1004,7 +1004,7 @@ func (cs *ConsensusState) enterPrecommit(height int, round int) {
cs.LockedRound = 0
cs.LockedBlock = nil
cs.LockedBlockParts = nil
types.FireEventUnlock(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventUnlock(cs.RoundStateEvent())
}
cs.signAddVote(types.VoteTypePrecommit, nil, types.PartSetHeader{})
return
@ -1016,7 +1016,7 @@ func (cs *ConsensusState) enterPrecommit(height int, round int) {
if cs.LockedBlock.HashesTo(blockID.Hash) {
cs.Logger.Info("enterPrecommit: +2/3 prevoted locked block. Relocking")
cs.LockedRound = round
types.FireEventRelock(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventRelock(cs.RoundStateEvent())
cs.signAddVote(types.VoteTypePrecommit, blockID.Hash, blockID.PartsHeader)
return
}
@ -1031,7 +1031,7 @@ func (cs *ConsensusState) enterPrecommit(height int, round int) {
cs.LockedRound = round
cs.LockedBlock = cs.ProposalBlock
cs.LockedBlockParts = cs.ProposalBlockParts
types.FireEventLock(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventLock(cs.RoundStateEvent())
cs.signAddVote(types.VoteTypePrecommit, blockID.Hash, blockID.PartsHeader)
return
}
@ -1047,7 +1047,7 @@ func (cs *ConsensusState) enterPrecommit(height int, round int) {
cs.ProposalBlock = nil
cs.ProposalBlockParts = types.NewPartSetFromHeader(blockID.PartsHeader)
}
types.FireEventUnlock(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventUnlock(cs.RoundStateEvent())
cs.signAddVote(types.VoteTypePrecommit, nil, types.PartSetHeader{})
}
@ -1191,21 +1191,19 @@ func (cs *ConsensusState) finalizeCommit(height int) {
// WAL replay for blocks with an #ENDHEIGHT
// As is, ConsensusState should not be started again
// until we successfully call ApplyBlock (ie. here or in Handshake after restart)
if cs.wal != nil {
cs.wal.Save(EndHeightMessage{uint64(height)})
}
cs.wal.Save(EndHeightMessage{uint64(height)})
fail.Fail() // XXX
// Create a copy of the state for staging
// and an event cache for txs
stateCopy := cs.state.Copy()
eventCache := types.NewEventCache(cs.evsw)
txEventBuffer := types.NewTxEventBuffer(cs.eventBus, block.NumTxs)
// Execute and commit the block, update and save the state, and update the mempool.
// All calls to the proxyAppConn come here.
// NOTE: the block.AppHash wont reflect these txs until the next block
err := stateCopy.ApplyBlock(eventCache, cs.proxyAppConn, block, blockParts.Header(), cs.mempool)
err := stateCopy.ApplyBlock(txEventBuffer, cs.proxyAppConn, block, blockParts.Header(), cs.mempool)
if err != nil {
cs.Logger.Error("Error on ApplyBlock. Did the application crash? Please restart tendermint", "err", err)
return
@ -1220,9 +1218,12 @@ func (cs *ConsensusState) finalizeCommit(height int) {
// * Fire before persisting state, in ApplyBlock
// * Fire on start up if we haven't written any new WAL msgs
// Both options mean we may fire more than once. Is that fine ?
types.FireEventNewBlock(cs.evsw, types.EventDataNewBlock{block})
types.FireEventNewBlockHeader(cs.evsw, types.EventDataNewBlockHeader{block.Header})
eventCache.Flush()
cs.eventBus.PublishEventNewBlock(types.EventDataNewBlock{block})
cs.eventBus.PublishEventNewBlockHeader(types.EventDataNewBlockHeader{block.Header})
err = txEventBuffer.Flush()
if err != nil {
cs.Logger.Error("Failed to flush event buffer", "err", err)
}
fail.Fail() // XXX
@ -1357,7 +1358,7 @@ func (cs *ConsensusState) addVote(vote *types.Vote, peerKey string) (added bool,
added, err = cs.LastCommit.AddVote(vote)
if added {
cs.Logger.Info(cmn.Fmt("Added to lastPrecommits: %v", cs.LastCommit.StringShort()))
types.FireEventVote(cs.evsw, types.EventDataVote{vote})
cs.eventBus.PublishEventVote(types.EventDataVote{vote})
// if we can skip timeoutCommit and have all the votes now,
if cs.config.SkipTimeoutCommit && cs.LastCommit.HasAll() {
@ -1375,7 +1376,7 @@ func (cs *ConsensusState) addVote(vote *types.Vote, peerKey string) (added bool,
height := cs.Height
added, err = cs.Votes.AddVote(vote, peerKey)
if added {
types.FireEventVote(cs.evsw, types.EventDataVote{vote})
cs.eventBus.PublishEventVote(types.EventDataVote{vote})
switch vote.Type {
case types.VoteTypePrevote:
@ -1393,7 +1394,7 @@ func (cs *ConsensusState) addVote(vote *types.Vote, peerKey string) (added bool,
cs.LockedRound = 0
cs.LockedBlock = nil
cs.LockedBlockParts = nil
types.FireEventUnlock(cs.evsw, cs.RoundStateEvent())
cs.eventBus.PublishEventUnlock(cs.RoundStateEvent())
}
}
if cs.Round <= vote.Round && prevotes.HasTwoThirdsAny() {