aboutsummaryrefslogtreecommitdiffstats
path: root/core/agreement-mgr.go
blob: d3a0af24fd917013bd3fbb04e5abf9b970acfe23 (plain) (blame)
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
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
// Copyright 2018 The dexon-consensus Authors
// This file is part of the dexon-consensus library.
//
// The dexon-consensus 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 dexon-consensus 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 dexon-consensus library. If not, see
// <http://www.gnu.org/licenses/>.

package core

import (
    "context"
    "errors"
    "sync"
    "time"

    "github.com/dexon-foundation/dexon-consensus/common"
    "github.com/dexon-foundation/dexon-consensus/core/types"
    "github.com/dexon-foundation/dexon-consensus/core/utils"
)

// Errors returned from BA modules
var (
    ErrPreviousRoundIsNotFinished = errors.New("previous round is not finished")
)

// genValidLeader generate a validLeader function for agreement modules.
func genValidLeader(
    mgr *agreementMgr) func(*types.Block) (bool, error) {
    return func(block *types.Block) (bool, error) {
        if block.Timestamp.After(time.Now()) {
            return false, nil
        }
        if err := mgr.lattice.SanityCheck(block); err != nil {
            if err == ErrRetrySanityCheckLater {
                return false, nil
            }
            return false, err
        }
        mgr.logger.Debug("Calling Application.VerifyBlock", "block", block)
        switch mgr.app.VerifyBlock(block) {
        case types.VerifyInvalidBlock:
            return false, ErrInvalidBlock
        case types.VerifyRetryLater:
            return false, nil
        default:
        }
        return true, nil
    }
}

type agreementMgrConfig struct {
    beginTime     time.Time
    numChains     uint32
    roundInterval time.Duration
    notarySetSize uint32
    lambdaBA      time.Duration
    crs           common.Hash
}

type baRoundSetting struct {
    chainID   uint32
    notarySet map[types.NodeID]struct{}
    agr       *agreement
    recv      *consensusBAReceiver
    ticker    Ticker
    crs       common.Hash
}

type agreementMgr struct {
    // TODO(mission): unbound Consensus instance from this module.
    con           *Consensus
    ID            types.NodeID
    app           Application
    gov           Governance
    network       Network
    logger        common.Logger
    cache         *utils.NodeSetCache
    auth          *Authenticator
    lattice       *Lattice
    ctx           context.Context
    lastEndTime   time.Time
    initRound     uint64
    configs       []*agreementMgrConfig
    baModules     []*agreement
    waitGroup     sync.WaitGroup
    pendingVotes  map[uint64][]*types.Vote
    pendingBlocks map[uint64][]*types.Block
    isRunning     bool

    // This lock should be used when attempting to:
    //  - add a new baModule.
    //  - remove all baModules when stopping. In this case, the cleaner need
    //    to wait for all routines runnning baModules finished.
    //  - access a method of baModule.
    //  - append a config from new round.
    // The routine running corresponding baModule, however, doesn't have to
    // acquire this lock.
    lock sync.RWMutex
}

func newAgreementMgr(con *Consensus, initRound uint64,
    initRoundBeginTime time.Time) *agreementMgr {
    return &agreementMgr{
        con:         con,
        ID:          con.ID,
        app:         con.app,
        gov:         con.gov,
        network:     con.network,
        logger:      con.logger,
        cache:       con.nodeSetCache,
        auth:        con.authModule,
        lattice:     con.lattice,
        ctx:         con.ctx,
        initRound:   initRound,
        lastEndTime: initRoundBeginTime,
    }
}

func (mgr *agreementMgr) getConfig(round uint64) *agreementMgrConfig {
    mgr.lock.RLock()
    defer mgr.lock.RUnlock()
    if round < mgr.initRound {
        panic(ErrRoundOutOfRange)
    }
    roundIndex := round - mgr.initRound
    if roundIndex >= uint64(len(mgr.configs)) {
        return nil
    }
    return mgr.configs[roundIndex]
}

func (mgr *agreementMgr) run() {
    mgr.lock.Lock()
    defer mgr.lock.Unlock()
    if mgr.isRunning {
        return
    }
    mgr.isRunning = true
    for i := uint32(0); i < uint32(len(mgr.baModules)); i++ {
        go mgr.runBA(mgr.initRound, i)
    }
}

func (mgr *agreementMgr) appendConfig(
    round uint64, config *types.Config, crs common.Hash) (err error) {
    mgr.lock.Lock()
    defer mgr.lock.Unlock()
    if round != uint64(len(mgr.configs))+mgr.initRound {
        return ErrRoundNotIncreasing
    }
    newConfig := &agreementMgrConfig{
        beginTime:     mgr.lastEndTime,
        numChains:     config.NumChains,
        roundInterval: config.RoundInterval,
        notarySetSize: config.NotarySetSize,
        lambdaBA:      config.LambdaBA,
        crs:           crs,
    }
    mgr.configs = append(mgr.configs, newConfig)
    mgr.lastEndTime = mgr.lastEndTime.Add(config.RoundInterval)
    // Create baModule for newly added chain.
    for i := uint32(len(mgr.baModules)); i < newConfig.numChains; i++ {
        // Prepare modules.
        recv := &consensusBAReceiver{
            consensus:     mgr.con,
            chainID:       i,
            restartNotary: make(chan bool, 1),
        }
        agrModule := newAgreement(
            mgr.con.ID,
            recv,
            newLeaderSelector(genValidLeader(mgr), mgr.logger),
            mgr.auth)
        // Hacky way to make agreement module self contained.
        recv.agreementModule = agrModule
        mgr.baModules = append(mgr.baModules, agrModule)
        if mgr.isRunning {
            go mgr.runBA(round, i)
        }
    }
    return nil
}

func (mgr *agreementMgr) processVote(v *types.Vote) error {
    v = v.Clone()
    mgr.lock.RLock()
    defer mgr.lock.RUnlock()
    if v.Position.ChainID >= uint32(len(mgr.baModules)) {
        mgr.logger.Error("Process vote for unknown chain to BA",
            "position", &v.Position,
            "baChain", len(mgr.baModules),
            "baRound", len(mgr.configs),
            "initRound", mgr.initRound)
        return utils.ErrInvalidChainID
    }
    return mgr.baModules[v.Position.ChainID].processVote(v)
}

func (mgr *agreementMgr) processBlock(b *types.Block) error {
    mgr.lock.RLock()
    defer mgr.lock.RUnlock()
    if b.Position.ChainID >= uint32(len(mgr.baModules)) {
        mgr.logger.Error("Process block for unknown chain to BA",
            "position", &b.Position,
            "baChain", len(mgr.baModules),
            "baRound", len(mgr.configs),
            "initRound", mgr.initRound)
        return utils.ErrInvalidChainID
    }
    return mgr.baModules[b.Position.ChainID].processBlock(b)
}

func (mgr *agreementMgr) processAgreementResult(
    result *types.AgreementResult) error {
    mgr.lock.RLock()
    defer mgr.lock.RUnlock()
    if result.Position.ChainID >= uint32(len(mgr.baModules)) {
        mgr.logger.Error("Process unknown result for unknown chain to BA",
            "position", &result.Position,
            "baChain", len(mgr.baModules),
            "baRound", len(mgr.configs),
            "initRound", mgr.initRound)
        return utils.ErrInvalidChainID
    }
    agreement := mgr.baModules[result.Position.ChainID]
    aID := agreement.agreementID()
    if isStop(aID) {
        return nil
    }
    if result.Position.Newer(&aID) {
        mgr.logger.Info("Syncing BA", "position", &result.Position)
        nodes, err := mgr.cache.GetNodeSet(result.Position.Round)
        if err != nil {
            return err
        }
        mgr.logger.Debug("Calling Network.PullBlocks for syncing BA",
            "hash", result.BlockHash)
        mgr.network.PullBlocks(common.Hashes{result.BlockHash})
        mgr.logger.Debug("Calling Governance.CRS", "round", result.Position.Round)
        crs := mgr.gov.CRS(result.Position.Round)
        nIDs := nodes.GetSubSet(
            int(mgr.gov.Configuration(result.Position.Round).NotarySetSize),
            types.NewNotarySetTarget(crs, result.Position.ChainID))
        for key := range result.Votes {
            agreement.processVote(&result.Votes[key])
        }
        agreement.restart(nIDs, result.Position, crs)
    }
    return nil
}

func (mgr *agreementMgr) stop() {
    // Stop all running agreement modules.
    func() {
        mgr.lock.Lock()
        defer mgr.lock.Unlock()
        for _, agr := range mgr.baModules {
            agr.stop()
        }
    }()
    // Block until all routines are done.
    mgr.waitGroup.Wait()
}

func (mgr *agreementMgr) runBA(initRound uint64, chainID uint32) {
    mgr.waitGroup.Add(1)
    defer mgr.waitGroup.Done()
    // Acquire agreement module.
    agr, recv := func() (*agreement, *consensusBAReceiver) {
        mgr.lock.RLock()
        defer mgr.lock.RUnlock()
        agr := mgr.baModules[chainID]
        return agr, agr.data.recv.(*consensusBAReceiver)
    }()
    // These are round based variables.
    var (
        currentRound uint64
        nextRound    = initRound
        setting      = baRoundSetting{
            chainID: chainID,
            agr:     agr,
            recv:    recv,
        }
        roundBeginTime time.Time
        roundEndTime   time.Time
        tickDuration   time.Duration
    )

    // Check if this routine needs to awake in this round and prepare essential
    // variables when yes.
    checkRound := func() (awake bool) {
        defer func() {
            currentRound = nextRound
            nextRound++
        }()
        // Wait until the configuartion for next round is ready.
        var config *agreementMgrConfig
        for {
            if config = mgr.getConfig(nextRound); config != nil {
                break
            } else {
                mgr.logger.Info("round is not ready", "round", nextRound)
                time.Sleep(1 * time.Second)
            }
        }
        // Set next checkpoint.
        roundBeginTime = config.beginTime
        roundEndTime = config.beginTime.Add(config.roundInterval)
        // Check if this chain handled by this routine included in this round.
        if chainID >= config.numChains {
            return false
        }
        // Check if this node in notary set of this chain in this round.
        nodeSet, err := mgr.cache.GetNodeSet(nextRound)
        if err != nil {
            panic(err)
        }
        setting.crs = config.crs
        setting.notarySet = nodeSet.GetSubSet(
            int(config.notarySetSize),
            types.NewNotarySetTarget(config.crs, chainID))
        _, awake = setting.notarySet[mgr.ID]
        // Setup ticker
        if tickDuration != config.lambdaBA {
            if setting.ticker != nil {
                setting.ticker.Stop()
            }
            setting.ticker = newTicker(mgr.gov, nextRound, TickerBA)
            tickDuration = config.lambdaBA
        }
        return
    }
Loop:
    for {
        select {
        case <-mgr.ctx.Done():
            break Loop
        default:
        }
        now := time.Now().UTC()
        if !checkRound() {
            if now.After(roundEndTime) {
                // That round is passed.
                continue Loop
            }
            // Sleep until next checkpoint.
            select {
            case <-mgr.ctx.Done():
                break Loop
            case <-time.After(roundEndTime.Sub(now)):
                continue Loop
            }
        }
        // Sleep until round begin. Here a biased round begin time would be
        // used instead of the one in config. The reason it to disperse the load
        // of fullnodes to verify confirmed blocks from each chain.
        if now.Before(pickBiasedTime(roundBeginTime, 4*tickDuration)) {
            select {
            case <-mgr.ctx.Done():
                break Loop
            case <-time.After(roundBeginTime.Sub(now)):
            }
            // Clean the tick channel after awake: the tick would be queued in
            // channel, thus the first few ticks would not tick on expected
            // interval.
            <-setting.ticker.Tick()
            <-setting.ticker.Tick()
        }
        // Run BA for this round.
        recv.round = currentRound
        recv.changeNotaryTime = roundEndTime
        recv.restartNotary <- false
        if err := mgr.baRoutineForOneRound(&setting); err != nil {
            mgr.logger.Error("BA routine failed",
                "error", err,
                "nodeID", mgr.ID,
                "chain", chainID)
            break Loop
        }
    }
}

func (mgr *agreementMgr) baRoutineForOneRound(
    setting *baRoundSetting) (err error) {
    agr := setting.agr
    recv := setting.recv
    oldPos := agr.agreementID()
Loop:
    for {
        select {
        case <-mgr.ctx.Done():
            break Loop
        default:
        }
        select {
        case newNotary := <-recv.restartNotary:
            if newNotary {
                // This round is finished.
                break Loop
            }
            var nextHeight uint64
            for {
                nextHeight, err = mgr.lattice.NextHeight(recv.round, setting.chainID)
                if err != nil {
                    mgr.logger.Debug("Error getting next height",
                        "error", err,
                        "round", recv.round,
                        "chainID", setting.chainID)
                    err = nil
                    nextHeight = oldPos.Height
                }
                if isStop(oldPos) || nextHeight == 0 {
                    break
                }
                if nextHeight > oldPos.Height {
                    break
                }
                mgr.logger.Debug("Lattice not ready!!!",
                    "old", &oldPos, "next", nextHeight)
                time.Sleep(100 * time.Millisecond)
            }
            nextPos := types.Position{
                Round:   recv.round,
                ChainID: setting.chainID,
                Height:  nextHeight,
            }
            oldPos = nextPos
            agr.restart(setting.notarySet, nextPos, setting.crs)
        default:
        }
        if agr.pullVotes() {
            pos := agr.agreementID()
            mgr.logger.Debug("Calling Network.PullVotes for syncing votes",
                "position", &pos)
            mgr.network.PullVotes(pos)
        }
        if err = agr.nextState(); err != nil {
            mgr.logger.Error("Failed to proceed to next state",
                "nodeID", mgr.ID.String(),
                "error", err)
            break Loop
        }
        for i := 0; i < agr.clocks(); i++ {
            // Priority select for agreement.done().
            select {
            case <-agr.done():
                continue Loop
            default:
            }
            select {
            case <-agr.done():
                continue Loop
            case <-setting.ticker.Tick():
            }
        }
    }
    return nil
}