simple (ticdc): support send all tables bootstrap message at changefeed start

cdc/api/v2/model.go

@@ -497,6 +497,10 @@ func (c *ReplicaConfig) toInternalReplicaConfigWithOriginConfig(
 		if c.Sink.SendBootstrapToAllPartition != nil {
 			res.Sink.SendBootstrapToAllPartition = util.AddressOf(*c.Sink.SendBootstrapToAllPartition)
 		}
+
+		if c.Sink.SendAllBootstrapAtStart != nil {
+			res.Sink.SendAllBootstrapAtStart = util.AddressOf(*c.Sink.SendAllBootstrapAtStart)
+		}
 	}
 	if c.Mounter != nil {
 		res.Mounter = &config.MounterConfig{
@@ -792,6 +796,10 @@ func ToAPIReplicaConfig(c *config.ReplicaConfig) *ReplicaConfig {
 			res.Sink.SendBootstrapToAllPartition = util.AddressOf(*cloned.Sink.SendBootstrapToAllPartition)
 		}
 
+		if cloned.Sink.SendAllBootstrapAtStart != nil {
+			res.Sink.SendAllBootstrapAtStart = util.AddressOf(*cloned.Sink.SendAllBootstrapAtStart)
+		}
+
 		if cloned.Sink.DebeziumDisableSchema != nil {
 			res.Sink.DebeziumDisableSchema = util.AddressOf(*cloned.Sink.DebeziumDisableSchema)
 		}
@@ -957,6 +965,7 @@ type SinkConfig struct {
 	SendBootstrapIntervalInSec       *int64              `json:"send_bootstrap_interval_in_sec,omitempty"`
 	SendBootstrapInMsgCount          *int32              `json:"send_bootstrap_in_msg_count,omitempty"`
 	SendBootstrapToAllPartition      *bool               `json:"send_bootstrap_to_all_partition,omitempty"`
+	SendAllBootstrapAtStart          *bool               `json:"send-all-bootstrap-at-start,omitempty"`
 	DebeziumDisableSchema            *bool               `json:"debezium_disable_schema,omitempty"`
 	DebeziumConfig                   *DebeziumConfig     `json:"debezium,omitempty"`
 	OpenProtocolConfig               *OpenProtocolConfig `json:"open,omitempty"`

cdc/api/v2/model_test.go

@@ -62,6 +62,7 @@ var defaultAPIConfig = &ReplicaConfig{
 		SendBootstrapIntervalInSec:       util.AddressOf(int64(120)),
 		SendBootstrapInMsgCount:          util.AddressOf(int32(10000)),
 		SendBootstrapToAllPartition:      util.AddressOf(true),
+		SendAllBootstrapAtStart:          util.AddressOf(false),
 		DebeziumDisableSchema:            util.AddressOf(false),
 		OpenProtocolConfig:               &OpenProtocolConfig{OutputOldValue: true},
 		DebeziumConfig:                   &DebeziumConfig{OutputOldValue: true},

cdc/entry/schema_test_helper.go

@@ -297,3 +297,8 @@ func (s *SchemaTestHelper) Close() {
 	s.domain.Close()
 	s.storage.Close() //nolint:errcheck
 }
+
+// SchemaStorage returns the schema storage
+func (s *SchemaTestHelper) SchemaStorage() SchemaStorage {
+	return s.schemaStorage
+}

cdc/owner/changefeed.go

@@ -716,7 +716,9 @@ LOOP2:
 		c.schema,
 		c.redoDDLMgr,
 		c.redoMetaMgr,
-		util.GetOrZero(cfInfo.Config.BDRMode))
+		util.GetOrZero(cfInfo.Config.BDRMode),
+		cfInfo.Config.Sink.ShouldSendAllBootstrapAtStart(),
+	)
 
 	// create scheduler
 	cfg := *c.cfg

cdc/owner/changefeed_test.go

@@ -89,7 +89,7 @@ type mockDDLSink struct {
 	// whether to record the DDL history, only for rename table
 	recordDDLHistory bool
 	// a slice of DDL history, only for rename table
-	ddlHistory []string
+	ddlHistory []*model.DDLEvent
 	mu         struct {
 		sync.Mutex
 		checkpointTs  model.Ts
@@ -117,7 +117,7 @@ func (m *mockDDLSink) emitDDLEvent(ctx context.Context, ddl *model.DDLEvent) (bo
 		}
 	}()
 	if m.recordDDLHistory {
-		m.ddlHistory = append(m.ddlHistory, ddl.Query)
+		m.ddlHistory = append(m.ddlHistory, ddl)
 	} else {
 		m.ddlHistory = nil
 	}
@@ -155,6 +155,13 @@ func (m *mockDDLSink) Barrier(ctx context.Context) error {
 	return nil
 }
 
+func (m *mockDDLSink) emitBootstrap(ctx context.Context, bootstrap *model.DDLEvent) error {
+	if m.recordDDLHistory {
+		m.ddlHistory = append(m.ddlHistory, bootstrap)
+	}
+	return nil
+}
+
 type mockScheduler struct {
 	currentTables []model.TableID
 }

cdc/owner/ddl_manager.go

@@ -130,6 +130,9 @@ type ddlManager struct {
 
 	BDRMode       bool
 	ddlResolvedTs model.Ts
+
+	shouldSendAllBootstrapAtStart bool
+	bootstraped                   bool
 }
 
 func newDDLManager(
@@ -143,6 +146,7 @@ func newDDLManager(
 	redoManager redo.DDLManager,
 	redoMetaManager redo.MetaManager,
 	bdrMode bool,
+	shouldSendAllBootstrapAtStart bool,
 ) *ddlManager {
 	log.Info("owner create ddl manager",
 		zap.String("namespace", changefeedID.Namespace),
@@ -152,19 +156,56 @@ func newDDLManager(
 		zap.Bool("bdrMode", bdrMode))
 
 	return &ddlManager{
-		changfeedID:     changefeedID,
-		ddlSink:         ddlSink,
-		filter:          filter,
-		ddlPuller:       ddlPuller,
-		schema:          schema,
-		redoDDLManager:  redoManager,
-		redoMetaManager: redoMetaManager,
-		startTs:         startTs,
-		checkpointTs:    checkpointTs,
-		ddlResolvedTs:   startTs,
-		BDRMode:         bdrMode,
-		pendingDDLs:     make(map[model.TableName][]*model.DDLEvent),
+		changfeedID:                   changefeedID,
+		ddlSink:                       ddlSink,
+		filter:                        filter,
+		ddlPuller:                     ddlPuller,
+		schema:                        schema,
+		redoDDLManager:                redoManager,
+		redoMetaManager:               redoMetaManager,
+		startTs:                       startTs,
+		checkpointTs:                  checkpointTs,
+		ddlResolvedTs:                 startTs,
+		BDRMode:                       bdrMode,
+		pendingDDLs:                   make(map[model.TableName][]*model.DDLEvent),
+		shouldSendAllBootstrapAtStart: shouldSendAllBootstrapAtStart,
+	}
+}
+
+func (m *ddlManager) checkAndSendBootstarpMsgs(ctx context.Context) (bool, error) {
+	if !m.shouldSendAllBootstrapAtStart || m.bootstraped {
+		return true, nil
+	}
+	start := time.Now()
+	defer func() {
+		log.Info("send bootstrap messages finished",
+			zap.Stringer("changefeed", m.changfeedID),
+			zap.Duration("cost", time.Since(start)))
+	}()
+	// Send bootstrap messages to downstream.
+	tableInfo, err := m.allTables(ctx)
+	if err != nil {
+		return false, errors.Trace(err)
+	}
+	log.Info("start to send bootstrap messages",
+		zap.Stringer("changefeed", m.changfeedID),
+		zap.Int("tables", len(tableInfo)))
+
+	for _, table := range tableInfo {
+		if table.TableInfo.IsView() {
+			continue
+		}
+		ddlEvent := &model.DDLEvent{
+			TableInfo:   table,
+			IsBootstrap: true,
+		}
+		err := m.ddlSink.emitBootstrap(ctx, ddlEvent)
+		if err != nil {
+			return false, errors.Trace(err)
+		}
 	}
+	m.bootstraped = true
+	return true, nil
 }
 
 // tick the ddlHandler, it does the following things:
@@ -183,6 +224,14 @@ func (m *ddlManager) tick(
 	m.justSentDDL = nil
 	m.checkpointTs = checkpointTs
 
+	ok, err := m.checkAndSendBootstarpMsgs(ctx)
+	if err != nil {
+		return nil, nil, errors.Trace(err)
+	}
+	if !ok {
+		return nil, nil, nil
+	}
+
 	currentTables, err := m.allTables(ctx)
 	if err != nil {
 		return nil, nil, errors.Trace(err)
@@ -483,8 +532,7 @@ func (m *ddlManager) barrier() *schedulepb.BarrierWithMinTs {
 	return barrier
 }
 
-// allTables returns all tables in the schema that
-// less or equal than the checkpointTs.
+// allTables returns all tables in the schema in current checkpointTs.
 func (m *ddlManager) allTables(ctx context.Context) ([]*model.TableInfo, error) {
 	if m.tableInfoCache == nil {
 		ts := m.getSnapshotTs()

cdc/owner/ddl_manager_test.go

@@ -50,7 +50,9 @@
 		schema,
 		redo.NewDisabledDDLManager(),
 		redo.NewDisabledMetaManager(),
-		false)
+		false,
+		false,
+	)
 	return res
 }
 
@@ -246,9 +248,9 @@
 	}
 	require.Len(t, mockDDLSink.ddlHistory, 2)
 	require.Equal(t, "RENAME TABLE `test1`.`tb1` TO `test2`.`tb10`",
-		mockDDLSink.ddlHistory[0])
+		mockDDLSink.ddlHistory[0].Query)
 	require.Equal(t, "RENAME TABLE `test2`.`tb2` TO `test1`.`tb20`",
-		mockDDLSink.ddlHistory[1])
+		mockDDLSink.ddlHistory[1].Query)
 
 	// mock all rename table statements have been done
 	mockDDLSink.resetDDLDone = false
@@ -459,3 +461,46 @@
 		require.Equal(t, c.ret, isGlobalDDL(c.ddl))
 	}
 }
+
+func TestCheckAndSendBootstarpMsgs(t *testing.T) {
+	helper := entry.NewSchemaTestHelper(t)
+	defer helper.Close()
+	ddl1 := helper.DDL2Event("create table test.tb1(id int primary key)")
+	ddl2 := helper.DDL2Event("create table test.tb2(id int primary key)")
+
+	ctx := context.Background()
+	dm := createDDLManagerForTest(t)
+	dm.schema = helper.SchemaStorage()
+	sendBootstrapIntervalInSec := int64(1)
+	sendBootstrapInMsgCount := int32(2)
+	protocol := "simple"
+	dm.sinkConfig.SendBootstrapIntervalInSec = &sendBootstrapIntervalInSec
+	dm.sinkConfig.SendBootstrapInMsgCount = &sendBootstrapInMsgCount
+	dm.sinkConfig.Protocol = &protocol
+	dm.startTs, dm.checkpointTs = ddl2.CommitTs, ddl2.CommitTs
+
+	mockDDLSink := dm.ddlSink.(*mockDDLSink)
+	mockDDLSink.recordDDLHistory = true
+
+	// do not send all bootstrap messages
+	sendAllBootstrapAtStart := false
+	dm.sinkConfig.SendAllBootstrapAtStart = &sendAllBootstrapAtStart
+	send, err := dm.checkAndSendBootstarpMsgs(ctx)
+	require.Nil(t, err)
+	require.True(t, send)
+	require.False(t, dm.bootstraped)
+	require.Equal(t, 0, len(mockDDLSink.ddlHistory))
+
+	// send all bootstrap messages -> tb1 and tb2
+	sendAllBootstrapAtStart = true
+	dm.sinkConfig.SendAllBootstrapAtStart = &sendAllBootstrapAtStart
+	send, err = dm.checkAndSendBootstarpMsgs(ctx)
+	require.Nil(t, err)
+	require.True(t, send)
+	require.True(t, dm.bootstraped)
+	require.Equal(t, 2, len(mockDDLSink.ddlHistory))
+	require.True(t, mockDDLSink.ddlHistory[0].IsBootstrap)
+	require.True(t, mockDDLSink.ddlHistory[1].IsBootstrap)
+	require.Equal(t, ddl1.TableInfo.TableName, mockDDLSink.ddlHistory[0].TableInfo.TableName)
+	require.Equal(t, ddl2.TableInfo.TableName, mockDDLSink.ddlHistory[1].TableInfo.TableName)
+}

cdc/owner/ddl_sink.go

@@ -54,6 +54,7 @@ type DDLSink interface {
 	// the caller of this function can call again and again until a true returned
 	emitDDLEvent(ctx context.Context, ddl *model.DDLEvent) (bool, error)
 	emitSyncPoint(ctx context.Context, checkpointTs uint64) error
+	emitBootstrap(ctx context.Context, bootstrap *model.DDLEvent) error
 	// close the ddlsink, cancel running goroutine.
 	close(ctx context.Context) error
 }
@@ -121,10 +122,6 @@ func ddlSinkInitializer(ctx context.Context, a *ddlSinkImpl) error {
 		return errors.Trace(err)
 	}
 	a.sink = s
-
-	if !util.GetOrZero(a.info.Config.EnableSyncPoint) {
-		return nil
-	}
 	return nil
 }
 
@@ -472,3 +469,10 @@ func (s *ddlSinkImpl) addSpecialComment(ddl *model.DDLEvent) (string, error) {
 
 	return result, nil
 }
+
+func (s *ddlSinkImpl) emitBootstrap(ctx context.Context, bootstrap *model.DDLEvent) error {
+	if err := s.makeSinkReady(ctx); err != nil {
+		return errors.Trace(err)
+	}
+	return s.sink.WriteDDLEvent(ctx, bootstrap)
+}

pkg/cmd/util/helper_test.go

@@ -213,6 +213,7 @@ func TestAndWriteExampleReplicaTOML(t *testing.T) {
 		SendBootstrapIntervalInSec:       util.AddressOf(int64(120)),
 		SendBootstrapInMsgCount:          util.AddressOf(int32(10000)),
 		SendBootstrapToAllPartition:      util.AddressOf(true),
+		SendAllBootstrapAtStart:          util.AddressOf(false),
 		DebeziumDisableSchema:            util.AddressOf(false),
 		OpenProtocol:                     &config.OpenProtocolConfig{OutputOldValue: true},
 		Debezium:                         &config.DebeziumConfig{OutputOldValue: true},
@@ -253,6 +254,7 @@ func TestAndWriteStorageSinkTOML(t *testing.T) {
 		SendBootstrapIntervalInSec:       util.AddressOf(int64(120)),
 		SendBootstrapInMsgCount:          util.AddressOf(int32(10000)),
 		SendBootstrapToAllPartition:      util.AddressOf(true),
+		SendAllBootstrapAtStart:          util.AddressOf(false),
 		DebeziumDisableSchema:            util.AddressOf(false),
 		OpenProtocol:                     &config.OpenProtocolConfig{OutputOldValue: true},
 		Debezium:                         &config.DebeziumConfig{OutputOldValue: true},

pkg/config/config_test_data.go

@@ -70,6 +70,7 @@ const (
     "send-bootstrap-interval-in-sec": 120,
     "send-bootstrap-in-msg-count": 10000,
     "send-bootstrap-to-all-partition": true,
+    "send-all-bootstrap-at-start": false,
     "debezium-disable-schema": false,
     "open": {
       "output-old-value": true
@@ -337,6 +338,7 @@ const (
     "send-bootstrap-interval-in-sec": 120,
     "send-bootstrap-in-msg-count": 10000,
     "send-bootstrap-to-all-partition": true,
+    "send-all-bootstrap-at-start": false,
     "debezium-disable-schema": false,
     "open": {
       "output-old-value": true
@@ -511,6 +513,7 @@ const (
     "send-bootstrap-interval-in-sec": 120,
     "send-bootstrap-in-msg-count": 10000,
     "send-bootstrap-to-all-partition": true,
+    "send-all-bootstrap-at-start": false,
     "debezium-disable-schema": false,
     "open": {
       "output-old-value": true

pkg/config/replica_config.go

@@ -77,6 +77,7 @@ var defaultReplicaConfig = &ReplicaConfig{
 		SendBootstrapIntervalInSec:       util.AddressOf(DefaultSendBootstrapIntervalInSec),
 		SendBootstrapInMsgCount:          util.AddressOf(DefaultSendBootstrapInMsgCount),
 		SendBootstrapToAllPartition:      util.AddressOf(DefaultSendBootstrapToAllPartition),
+		SendAllBootstrapAtStart:          util.AddressOf(DefaultSendAllBootstrapAtStart),
 		DebeziumDisableSchema:            util.AddressOf(false),
 		OpenProtocol:                     &OpenProtocolConfig{OutputOldValue: true},
 		Debezium:                         &DebeziumConfig{OutputOldValue: true},

pkg/config/sink.go

@@ -89,6 +89,9 @@ const (
 	// DefaultSendBootstrapToAllPartition is the default value of
 	// whether to send bootstrap message to all partitions.
 	DefaultSendBootstrapToAllPartition = true
+	// DefaultSendAllBootstrapAtStart is the default value of whether
+	// to send all tables bootstrap message at changefeed start.
+	DefaultSendAllBootstrapAtStart = false
 
 	// DefaultMaxReconnectToPulsarBroker is the default max reconnect times to pulsar broker.
 	// The pulsar client uses an exponential backoff with jitter to reconnect to the broker.
@@ -188,7 +191,8 @@ type SinkConfig struct {
 	// If set to false, bootstrap message will only be sent to the first partition of each topic.
 	// Default value is true.
 	SendBootstrapToAllPartition *bool `toml:"send-bootstrap-to-all-partition" json:"send-bootstrap-to-all-partition,omitempty"`
-
+	// SendAllBootstrapAtStart determines whether to send all tables bootstrap message at changefeed start.
+	SendAllBootstrapAtStart *bool `toml:"send-all-bootstrap-at-start" json:"send-all-bootstrap-at-start,omitempty"`
 	// Debezium only. Whether schema should be excluded in the output.
 	DebeziumDisableSchema *bool `toml:"debezium-disable-schema" json:"debezium-disable-schema,omitempty"`
 
@@ -227,6 +231,16 @@ func (s *SinkConfig) ShouldSendBootstrapMsg() bool {
 		util.GetOrZero(s.SendBootstrapInMsgCount) > 0
 }
 
+// ShouldSendAllBootstrapAtStart returns whether the should send all bootstrap message at changefeed start.
+func (s *SinkConfig) ShouldSendAllBootstrapAtStart() bool {
+	if s == nil {
+		return false
+	}
+	should := s.ShouldSendBootstrapMsg() && util.GetOrZero(s.SendAllBootstrapAtStart)
+	log.Info("should send all bootstrap at start", zap.Bool("should", should))
+	return should
+}
+
 // CSVConfig defines a series of configuration items for csv codec.
 type CSVConfig struct {
 	// delimiter between fields, it can be 1 character or at most 2 characters