1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
|
func (bcR *BlockchainReactor) poolRoutine() {
trySyncTicker := time.NewTicker(trySyncIntervalMS * time.Millisecond)
statusUpdateTicker := time.NewTicker(statusUpdateIntervalSeconds * time.Second)
switchToConsensusTicker := time.NewTicker(switchToConsensusIntervalSeconds * time.Second)
blocksSynced := 0
chainID := bcR.initialState.ChainID
state := bcR.initialState
lastHundred := time.Now()
lastRate := 0.0
didProcessCh := make(chan struct{}, 1)
FOR_LOOP:
for {
select {
case request := <-bcR.requestsCh:
peer := bcR.Switch.Peers().Get(request.PeerID)
if peer == nil {
continue FOR_LOOP // Peer has since been disconnected.
}
msgBytes := cdc.MustMarshalBinaryBare(&bcBlockRequestMessage{request.Height})
queued := peer.TrySend(BlockchainChannel, msgBytes)
if !queued {
// We couldn't make the request, send-queue full.
// The pool handles timeouts, just let it go.
continue FOR_LOOP
}
case err := <-bcR.errorsCh:
peer := bcR.Switch.Peers().Get(err.peerID)
if peer != nil {
bcR.Switch.StopPeerForError(peer, err)
}
case <-statusUpdateTicker.C:
// ask for status updates
go bcR.BroadcastStatusRequest() // nolint: errcheck
case <-switchToConsensusTicker.C:
height, numPending, lenRequesters := bcR.pool.GetStatus()
outbound, inbound, _ := bcR.Switch.NumPeers()
bcR.Logger.Debug("Consensus ticker", "numPending", numPending, "total", lenRequesters,
"outbound", outbound, "inbound", inbound)
if bcR.pool.IsCaughtUp() {
bcR.Logger.Info("Time to switch to consensus reactor!", "height", height)
bcR.pool.Stop()
conR, ok := bcR.Switch.Reactor("CONSENSUS").(consensusReactor)
if ok {
conR.SwitchToConsensus(state, blocksSynced)
} else {
// should only happen during testing
}
break FOR_LOOP
}
case <-trySyncTicker.C: // chan time
select {
case didProcessCh <- struct{}{}:
default:
}
case <-didProcessCh:
// NOTE: It is a subtle mistake to process more than a single block
// at a time (e.g. 10) here, because we only TrySend 1 request per
// loop. The ratio mismatch can result in starving of blocks, a
// sudden burst of requests and responses, and repeat.
// Consequently, it is better to split these routines rather than
// coupling them as it's written here. TODO uncouple from request
// routine.
// See if there are any blocks to sync.
first, second := bcR.pool.PeekTwoBlocks()
//bcR.Logger.Info("TrySync peeked", "first", first, "second", second)
if first == nil || second == nil {
// We need both to sync the first block.
continue FOR_LOOP
} else {
// Try again quickly next loop.
didProcessCh <- struct{}{}
}
firstParts := first.MakePartSet(types.BlockPartSizeBytes)
firstPartsHeader := firstParts.Header()
firstID := types.BlockID{first.Hash(), firstPartsHeader}
// Finally, verify the first block using the second's commit
// NOTE: we can probably make this more efficient, but note that calling
// first.Hash() doesn't verify the tx contents, so MakePartSet() is
// currently necessary.
err := state.Validators.VerifyCommit(
chainID, firstID, first.Height, second.LastCommit)
if err != nil {
bcR.Logger.Error("Error in validation", "err", err)
peerID := bcR.pool.RedoRequest(first.Height)
peer := bcR.Switch.Peers().Get(peerID)
if peer != nil {
// NOTE: we've already removed the peer's request, but we
// still need to clean up the rest.
bcR.Switch.StopPeerForError(peer, fmt.Errorf("BlockchainReactor validation error: %v", err))
}
peerID2 := bcR.pool.RedoRequest(second.Height)
peer2 := bcR.Switch.Peers().Get(peerID2)
if peer2 != nil && peer2 != peer {
// NOTE: we've already removed the peer's request, but we
// still need to clean up the rest.
bcR.Switch.StopPeerForError(peer2, fmt.Errorf("BlockchainReactor validation error: %v", err))
}
continue FOR_LOOP
} else {
bcR.pool.PopRequest()
// TODO: batch saves so we dont persist to disk every block
bcR.store.SaveBlock(first, firstParts, second.LastCommit)
// TODO: same thing for app - but we would need a way to
// get the hash without persisting the state
var err error
state, err = bcR.blockExec.ApplyBlock(state, firstID, first)
if err != nil {
// TODO This is bad, are we zombie?
cmn.PanicQ(fmt.Sprintf("Failed to process committed block (%d:%X): %v",
first.Height, first.Hash(), err))
}
blocksSynced++
if blocksSynced%100 == 0 {
lastRate = 0.9*lastRate + 0.1*(100/time.Since(lastHundred).Seconds())
bcR.Logger.Info("Fast Sync Rate", "height", bcR.pool.height,
"max_peer_height", bcR.pool.MaxPeerHeight(), "blocks/s", lastRate)
lastHundred = time.Now()
}
}
continue FOR_LOOP
case <-bcR.Quit():
break FOR_LOOP
}
}
}
|
