forked from hyperledger/fabric
/
node.go
359 lines (296 loc) · 9.49 KB
/
node.go
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
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
/*
Copyright IBM Corp. All Rights Reserved.
SPDX-License-Identifier: Apache-2.0
*/
package etcdraft
import (
"context"
"crypto/sha256"
"sync"
"sync/atomic"
"time"
"code.cloudfoundry.org/clock"
"github.com/ZihuaZhang/fabric-protos-go/orderer"
"github.com/ZihuaZhang/fabric-protos-go/orderer/etcdraft"
"github.com/ZihuaZhang/fabric/common/flogging"
"github.com/ZihuaZhang/fabric/protoutil"
"github.com/golang/protobuf/proto"
"github.com/pkg/errors"
"go.etcd.io/etcd/raft/v3"
"go.etcd.io/etcd/raft/v3/raftpb"
)
var (
ErrChainHalting = errors.New("chain halting is in progress")
ErrNoAvailableLeaderCandidate = errors.New("leadership transfer failed to identify transferee")
ErrTimedOutLeaderTransfer = errors.New("leadership transfer timed out")
ErrNoLeader = errors.New("no leader")
)
type node struct {
chainID string
logger *flogging.FabricLogger
metrics *Metrics
unreachableLock sync.RWMutex
unreachable map[uint64]struct{}
tracker *Tracker
storage *RaftStorage
config *raft.Config
confState atomic.Value // stores raft ConfState
rpc RPC
chain *Chain
tickInterval time.Duration
clock clock.Clock
metadata *etcdraft.BlockMetadata
leaderChangeSubscription atomic.Value
raft.Node
}
func (n *node) start(fresh, join bool) {
raftPeers := RaftPeers(n.metadata.ConsenterIds)
n.logger.Debugf("Starting raft node: #peers: %v", len(raftPeers))
var campaign bool
if fresh {
if join {
n.logger.Info("Starting raft node to join an existing channel")
n.Node = raft.RestartNode(n.config)
} else {
n.logger.Info("Starting raft node as part of a new channel")
// determine the node to start campaign by selecting the node with ID equals to:
// hash(channelID) % cluster_size + 1
sha := sha256.Sum256([]byte(n.chainID))
number, _ := proto.DecodeVarint(sha[24:])
if n.config.ID == number%uint64(len(raftPeers))+1 {
campaign = true
}
n.Node = raft.StartNode(n.config, raftPeers)
}
} else {
n.logger.Info("Restarting raft node")
n.Node = raft.RestartNode(n.config)
}
go n.run(campaign)
}
func (n *node) run(campaign bool) {
electionTimeout := n.tickInterval.Seconds() * float64(n.config.ElectionTick)
halfElectionTimeout := electionTimeout / 2
raftTicker := n.clock.NewTicker(n.tickInterval)
if s := n.storage.Snapshot(); !raft.IsEmptySnap(s) {
n.chain.snapC <- &s
}
elected := make(chan struct{})
if campaign {
n.logger.Infof("This node is picked to start campaign")
go func() {
// Attempt campaign every two HeartbeatTimeout elapses, until leader is present - either this
// node successfully claims leadership, or another leader already existed when this node starts.
// We could do this more lazily and exit proactive campaign once transitioned to Candidate state
// (not PreCandidate because other nodes might not have started yet, in which case PreVote
// messages are dropped at recipients). But there is no obvious reason (for now) to be lazy.
//
// 2*HeartbeatTick is used to avoid excessive campaign when network latency is significant and
// Raft term keeps advancing in this extreme case.
campaignTicker := n.clock.NewTicker(n.tickInterval * time.Duration(n.config.HeartbeatTick) * 2)
defer campaignTicker.Stop()
for {
select {
case <-campaignTicker.C():
n.Campaign(context.TODO())
case <-elected:
return
case <-n.chain.doneC:
return
}
}
}()
}
for {
select {
case <-raftTicker.C():
// grab raft Status before ticking it, so `RecentActive` attributes
// are not reset yet.
status := n.Status()
n.Tick()
n.tracker.Check(&status)
case rd := <-n.Ready():
startStoring := n.clock.Now()
if err := n.storage.Store(rd.Entries, rd.HardState, rd.Snapshot); err != nil {
n.logger.Panicf("Failed to persist etcd/raft data: %s", err)
}
duration := n.clock.Since(startStoring).Seconds()
n.metrics.DataPersistDuration.Observe(duration)
if duration > halfElectionTimeout {
n.logger.Warningf("WAL sync took %v seconds and the network is configured to start elections after %v seconds. Your disk is too slow and may cause loss of quorum and trigger leadership election.", duration, electionTimeout)
}
if !raft.IsEmptySnap(rd.Snapshot) {
n.chain.snapC <- &rd.Snapshot
}
// skip empty apply
if len(rd.CommittedEntries) != 0 || rd.SoftState != nil {
n.maybeSyncWAL(rd.CommittedEntries)
n.chain.applyC <- apply{rd.CommittedEntries, rd.SoftState}
}
if campaign && rd.SoftState != nil {
leader := atomic.LoadUint64(&rd.SoftState.Lead) // etcdraft requires atomic access to this var
if leader != raft.None {
n.logger.Infof("Leader %d is present, quit campaign", leader)
campaign = false
close(elected)
}
}
n.Advance()
// TODO(jay_guo) leader can write to disk in parallel with replicating
// to the followers and them writing to their disks. Check 10.2.1 in thesis
n.send(rd.Messages)
case <-n.chain.haltC:
raftTicker.Stop()
n.Stop()
n.storage.Close()
n.logger.Infof("Raft node stopped")
close(n.chain.doneC) // close after all the artifacts are closed
return
}
}
}
func (n *node) send(msgs []raftpb.Message) {
n.unreachableLock.RLock()
defer n.unreachableLock.RUnlock()
for _, msg := range msgs {
if msg.To == 0 {
continue
}
status := raft.SnapshotFinish
// Replace node list in snapshot with CURRENT node list in cluster.
if msg.Type == raftpb.MsgSnap {
state := n.confState.Load()
if state != nil {
msg.Snapshot.Metadata.ConfState = *state.(*raftpb.ConfState)
}
}
msgBytes := protoutil.MarshalOrPanic(&msg)
err := n.rpc.SendConsensus(msg.To, &orderer.ConsensusRequest{Channel: n.chainID, Payload: msgBytes})
if err != nil {
n.ReportUnreachable(msg.To)
n.logSendFailure(msg.To, err)
status = raft.SnapshotFailure
} else if _, ok := n.unreachable[msg.To]; ok {
n.logger.Infof("Successfully sent StepRequest to %d after failed attempt(s)", msg.To)
delete(n.unreachable, msg.To)
}
if msg.Type == raftpb.MsgSnap {
n.ReportSnapshot(msg.To, status)
}
}
}
func (n *node) maybeSyncWAL(entries []raftpb.Entry) {
allNormal := true
for _, entry := range entries {
if entry.Type == raftpb.EntryNormal {
continue
}
allNormal = false
}
if allNormal {
return
}
if err := n.storage.Sync(); err != nil {
n.logger.Errorf("Failed to sync raft log, error: %s", err)
}
}
// abdicateLeadership picks a node that is recently active, and attempts to transfer leadership to it.
// Blocks until leadership transfer happens or when a timeout expires.
// Returns error upon failure.
func (n *node) abdicateLeadership() error {
start := time.Now()
defer func() {
n.logger.Infof("abdicateLeader took %v", time.Since(start))
}()
status := n.Status()
if status.Lead == raft.None {
n.logger.Warn("No leader, cannot transfer leadership")
return ErrNoLeader
}
if status.Lead != n.config.ID {
n.logger.Warn("Leader has changed since asked to transfer leadership")
return nil
}
// register to leader changes
notifyC, unsubscribe := n.subscribeToLeaderChange()
defer unsubscribe()
var transferee uint64
for id, pr := range status.Progress {
if id == status.ID {
continue // skip self
}
if pr.RecentActive && !pr.IsPaused() {
transferee = id
break
}
n.logger.Debugf("Node %d is not qualified as transferee because it's either paused or not active", id)
}
if transferee == raft.None {
n.logger.Errorf("No follower is qualified as transferee, abort leader transfer")
return ErrNoAvailableLeaderCandidate
}
n.logger.Infof("Transferring leadership to %d", transferee)
timeToWait := time.Duration(n.config.ElectionTick) * n.tickInterval
n.logger.Infof("Will wait %v time to abdicate", timeToWait)
ctx, cancel := context.WithTimeout(context.TODO(), timeToWait)
defer cancel()
n.TransferLeadership(ctx, status.ID, transferee)
timer := n.clock.NewTimer(timeToWait)
defer timer.Stop()
for {
select {
case <-timer.C():
n.logger.Warn("Leader transfer timed out")
return ErrTimedOutLeaderTransfer
case l := <-notifyC:
n.logger.Infof("Leader has been transferred from %d to %d", n.config.ID, l)
return nil
case <-n.chain.doneC:
n.logger.Infof("Returning early because chain is halting")
return ErrChainHalting
}
}
}
func (n *node) subscribeToLeaderChange() (chan uint64, func()) {
notifyC := make(chan uint64, 1)
subscriptionActive := uint32(1)
unsubscribe := func() {
atomic.StoreUint32(&subscriptionActive, 0)
}
subscription := func(leader uint64) {
if atomic.LoadUint32(&subscriptionActive) == 0 {
return
}
if leader != n.config.ID {
select {
case notifyC <- leader:
default:
// In case notifyC is full
}
}
}
n.leaderChangeSubscription.Store(subscription)
return notifyC, unsubscribe
}
func (n *node) logSendFailure(dest uint64, err error) {
if _, ok := n.unreachable[dest]; ok {
n.logger.Debugf("Failed to send StepRequest to %d, because: %s", dest, err)
return
}
n.logger.Errorf("Failed to send StepRequest to %d, because: %s", dest, err)
n.unreachable[dest] = struct{}{}
}
func (n *node) takeSnapshot(index uint64, cs raftpb.ConfState, data []byte) {
if err := n.storage.TakeSnapshot(index, cs, data); err != nil {
n.logger.Errorf("Failed to create snapshot at index %d: %s", index, err)
}
}
func (n *node) lastIndex() uint64 {
i, _ := n.storage.ram.LastIndex()
return i
}
func (n *node) ApplyConfChange(cc raftpb.ConfChange) *raftpb.ConfState {
state := n.Node.ApplyConfChange(cc)
n.confState.Store(state)
return state
}