-
Notifications
You must be signed in to change notification settings - Fork 22
/
sqlstore_broker.go
401 lines (348 loc) · 13 KB
/
sqlstore_broker.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
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
// Copyright (C) 2023 Gobalsky Labs Limited
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as
// published by the Free Software Foundation, either version 3 of the
// License, or (at your option) any later version.
//
// This program 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 Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
package broker
import (
"context"
"fmt"
"reflect"
"time"
"code.vegaprotocol.io/vega/core/events"
"code.vegaprotocol.io/vega/datanode/entities"
"code.vegaprotocol.io/vega/datanode/metrics"
"code.vegaprotocol.io/vega/logging"
"github.com/pkg/errors"
)
type SQLBrokerSubscriber interface {
SetVegaTime(vegaTime time.Time)
Flush(ctx context.Context) error
Push(ctx context.Context, val events.Event) error
Types() []events.Type
}
type SQLStoreEventBroker interface {
Receive(ctx context.Context) error
}
type TransactionManager interface {
WithConnection(ctx context.Context) (context.Context, error)
WithTransaction(ctx context.Context) (context.Context, error)
Commit(ctx context.Context) error
Rollback(ctx context.Context) error
RefreshMaterializedViews(ctx context.Context) error
}
type BlockStore interface {
Add(ctx context.Context, b entities.Block) error
GetLastBlock(ctx context.Context) (entities.Block, error)
}
type ProtocolUpgradeHandler interface {
OnProtocolUpgradeEvent(ctx context.Context, chainID string, lastCommittedBlockHeight int64) error
GetProtocolUpgradeStarted() bool
}
const (
slowTimeUpdateThreshold = 2 * time.Second
)
// SQLStoreBroker : push events to each subscriber with a single go routine across all types.
type SQLStoreBroker struct {
config Config
log *logging.Logger
subscribers []SQLBrokerSubscriber
typeToSubs map[events.Type][]SQLBrokerSubscriber
eventSource EventReceiver
transactionManager TransactionManager
blockStore BlockStore
onBlockCommitted func(ctx context.Context, chainId string, lastCommittedBlockHeight int64, snapshotTaken bool)
protocolUpdateHandler ProtocolUpgradeHandler
chainID string
lastBlock *entities.Block
slowTimeUpdateTicker *time.Ticker
receivedProtocolUpgradeEvent bool
snapshotTaken bool
}
func NewSQLStoreBroker(
log *logging.Logger,
config Config,
chainID string,
eventsource EventReceiver,
transactionManager TransactionManager,
blockStore BlockStore,
onBlockCommitted func(ctx context.Context, chainId string, lastCommittedBlockHeight int64, snapshotTaken bool),
protocolUpdateHandler ProtocolUpgradeHandler,
subs []SQLBrokerSubscriber,
) *SQLStoreBroker {
b := &SQLStoreBroker{
config: config,
log: log.Named("sqlstore-broker"),
subscribers: subs,
typeToSubs: map[events.Type][]SQLBrokerSubscriber{},
eventSource: eventsource,
transactionManager: transactionManager,
blockStore: blockStore,
chainID: chainID,
onBlockCommitted: onBlockCommitted,
protocolUpdateHandler: protocolUpdateHandler,
slowTimeUpdateTicker: time.NewTicker(slowTimeUpdateThreshold),
}
for _, s := range subs {
for _, evtType := range s.Types() {
b.typeToSubs[evtType] = append(b.typeToSubs[evtType], s)
}
}
return b
}
func (b *SQLStoreBroker) Receive(ctx context.Context) error {
if err := b.eventSource.Listen(); err != nil {
return err
}
receiveCh, errCh := b.eventSource.Receive(ctx)
nextBlock, err := b.waitForFirstBlock(ctx, errCh, receiveCh)
if err != nil {
return err
}
dbContext, err := b.transactionManager.WithConnection(context.Background())
if err != nil {
return err
}
for {
if nextBlock, err = b.processBlock(ctx, dbContext, nextBlock, receiveCh, errCh); err != nil {
return err
}
b.onBlockCommitted(ctx, b.chainID, b.lastBlock.Height, b.snapshotTaken)
if b.receivedProtocolUpgradeEvent {
return b.protocolUpdateHandler.OnProtocolUpgradeEvent(ctx, b.chainID, b.lastBlock.Height)
}
}
}
// waitForFirstBlock processes all events until a new block is encountered and returns the new block. A 'new' block is one for which
// events have not already been processed by this datanode.
func (b *SQLStoreBroker) waitForFirstBlock(ctx context.Context, errCh <-chan error, receiveCh <-chan events.Event) (*entities.Block, error) {
lastProcessedBlock, err := b.blockStore.GetLastBlock(ctx)
if err == nil {
b.log.Infof("waiting for first unprocessed block, last processed block height: %d", lastProcessedBlock.Height)
} else if errors.Is(err, entities.ErrNotFound) {
lastProcessedBlock = entities.Block{
VegaTime: time.Time{},
// TODO: This is making the assumption that the first block will be height 1. This is not necessarily true.
// The node can start at any time given to Tendermint through the genesis file.
Height: 0,
Hash: nil,
}
} else {
return nil, err
}
var beginBlock entities.BeginBlockEvent
for {
select {
case <-ctx.Done():
return nil, ctx.Err()
case err = <-errCh:
return nil, err
case e := <-receiveCh:
if e.Type() == events.BeginBlockEvent {
beginBlock = e.(entities.BeginBlockEvent)
metrics.EventCounterInc(beginBlock.Type().String())
if beginBlock.BlockNr() > lastProcessedBlock.Height+1 {
return nil, fmt.Errorf("block height on begin block, %d, is too high, the height of the last processed block is %d",
beginBlock.BlockNr(), lastProcessedBlock.Height)
}
if beginBlock.BlockNr() > lastProcessedBlock.Height {
b.log.Info("first unprocessed block received, starting block processing")
return entities.BlockFromBeginBlock(beginBlock)
}
}
}
}
}
// processBlock processes all events in the current block up to the next time update. The next time block is returned when processing of the block is done.
func (b *SQLStoreBroker) processBlock(ctx context.Context, dbContext context.Context, block *entities.Block, eventsCh <-chan events.Event, errCh <-chan error) (*entities.Block, error) {
metrics.BlockCounterInc()
metrics.SetBlockHeight(float64(block.Height))
blockTimer := blockTimer{}
blockTimer.startTimer()
defer func() {
blockTimer.stopTimer()
metrics.AddBlockHandlingTime(blockTimer.duration)
}()
for _, subscriber := range b.subscribers {
subscriber.SetVegaTime(block.VegaTime)
}
// Don't use our parent context as a parent of the database operation; if we get cancelled
// by e.g. a shutdown request then let the last database operation complete.
var blockCtx context.Context
var cancel context.CancelFunc
blockCtx, cancel = context.WithCancel(dbContext)
defer cancel()
blockCtx, err := b.transactionManager.WithTransaction(blockCtx)
defer b.transactionManager.Rollback(blockCtx)
if err != nil {
return nil, fmt.Errorf("failed to add transaction to context:%w", err)
}
if err = b.addBlock(blockCtx, block); err != nil {
return nil, fmt.Errorf("failed to add block:%w", err)
}
defer b.slowTimeUpdateTicker.Stop()
b.snapshotTaken = false
betweenBlocks := false
refreshMaterializedViews := false
for {
// Do a pre-check on ctx.Done() since select() cases are randomized, this reduces
// the number of things we'll keep trying to handle after we are cancelled.
select {
case <-ctx.Done():
return nil, ctx.Err()
default:
}
blockTimer.stopTimer()
select {
case <-ctx.Done():
return nil, ctx.Err()
case err = <-errCh:
return nil, err
case <-b.slowTimeUpdateTicker.C:
b.log.Warningf("slow time update detected, time between checks %v, block height: %d, total block processing time: %v", slowTimeUpdateThreshold,
block.Height, blockTimer.duration)
case e := <-eventsCh:
if b.protocolUpdateHandler.GetProtocolUpgradeStarted() {
return nil, errors.New("received event after protocol upgrade started")
}
if b.config.Level.Level == logging.DebugLevel {
b.log.Debug("received event", logging.String("type", e.Type().String()), logging.String("trace-id", e.TraceID()))
}
metrics.EventCounterInc(e.Type().String())
blockTimer.startTimer()
switch e.Type() {
case events.EndBlockEvent:
err = b.flushAllSubscribers(blockCtx)
if err != nil {
return nil, err
}
err = b.transactionManager.Commit(blockCtx)
if err != nil {
return nil, fmt.Errorf("failed to commit transactional context:%w", err)
}
b.slowTimeUpdateTicker.Reset(slowTimeUpdateThreshold)
betweenBlocks = true
if err = b.handleEvent(blockCtx, e); err != nil {
return nil, err
}
// at the end of the block, if we have had an epoch event in that block then we should have received
// statistics that were updated and reported only at the end of an epoch. The refreshMaterialized flag
// should have been set by the EpochUpdate event before this EndBlockEvent was received
// so we need to call the refresh materialized views function here.
if refreshMaterializedViews {
// We need to refresh the materialized views as we have reached the end of an epoch
err = b.transactionManager.RefreshMaterializedViews(blockCtx)
if err != nil {
return nil, fmt.Errorf("failed to refresh materialized views:%w", err)
}
refreshMaterializedViews = false
}
case events.BeginBlockEvent:
beginBlock := e.(entities.BeginBlockEvent)
return entities.BlockFromBeginBlock(beginBlock)
case events.CoreSnapshotEvent:
// if a snapshot is taken on a protocol upgrade block, we want it to be taken synchronously as part of handling of protocol upgrade
b.snapshotTaken = !e.StreamMessage().GetCoreSnapshotEvent().ProtocolUpgradeBlock
if err = b.handleEvent(blockCtx, e); err != nil {
return nil, err
}
case events.ProtocolUpgradeStartedEvent:
b.receivedProtocolUpgradeEvent = true
// we've received a protocol upgrade event which is the last event core will have sent out
// so we can leave now
return nil, nil
case events.EpochUpdate:
// We have received an epoch event in this block, so we set a flag that will indicate that we should
// refresh any materialized views that need to be refreshed after receiving data that is only sent
// once an epoch.
refreshMaterializedViews = true
// We want the default block to execute after we have done this so we fall through to the default case
// DANGER WILL ROBINSON!!! Make sure you don't add any code here that will prevent the fallthrough
// or add another case statement that will prevent the fallthrough to the default case
fallthrough
default:
if betweenBlocks {
// we should only be receiving a BeginBlockEvent immediately after an EndBlockEvent
panic(fmt.Errorf("received event %s between end block and begin block", e.Type().String()))
}
if err = b.handleEvent(blockCtx, e); err != nil {
return nil, err
}
}
}
}
}
func (b *SQLStoreBroker) flushAllSubscribers(blockCtx context.Context) error {
for _, subscriber := range b.subscribers {
subName := reflect.TypeOf(subscriber).Elem().Name()
timer := metrics.NewTimeCounter(subName)
err := subscriber.Flush(blockCtx)
timer.FlushTimeCounterAdd()
if err != nil {
return fmt.Errorf("failed to flush subscriber:%w", err)
}
}
return nil
}
func (b *SQLStoreBroker) addBlock(ctx context.Context, block *entities.Block) error {
// At startup we get time updates that have the same time to microsecond precision which causes
// a primary key restraint failure, this code is to handle this scenario
if b.lastBlock == nil || !block.VegaTime.Equal(b.lastBlock.VegaTime) {
b.lastBlock = block
err := b.blockStore.Add(ctx, *block)
if err != nil {
return errors.Wrap(err, "failed to add block")
}
}
return nil
}
func (b *SQLStoreBroker) handleEvent(ctx context.Context, e events.Event) error {
if err := checkChainID(b.chainID, e.ChainID()); err != nil {
return err
}
if subs, ok := b.typeToSubs[e.Type()]; ok {
for _, sub := range subs {
if err := b.push(ctx, sub, e); err != nil {
return err
}
}
}
return nil
}
func (b *SQLStoreBroker) push(ctx context.Context, sub SQLBrokerSubscriber, e events.Event) error {
subName := reflect.TypeOf(sub).Elem().Name()
timer := metrics.NewTimeCounter("sql", subName, e.Type().String())
err := sub.Push(ctx, e)
timer.EventTimeCounterAdd()
if err != nil {
errMsg := fmt.Sprintf("failed to process event %v error:%+v", e.StreamMessage(), err)
b.log.Error(errMsg)
if b.config.PanicOnError {
return err
}
}
return nil
}
type blockTimer struct {
duration time.Duration
startTime *time.Time
}
func (t *blockTimer) startTimer() {
now := time.Now()
t.startTime = &now
}
func (t *blockTimer) stopTimer() {
if t.startTime != nil {
t.duration = t.duration + time.Since(*t.startTime)
t.startTime = nil
}
}