processor: fix a bug that will cause processor Tick get stuck when downstream is Kafka

cdc/processor/sinkmanager/manager.go

@@ -33,7 +33,6 @@ import (
 	"github.com/pingcap/tiflow/cdc/sink/dmlsink/factory"
 	tablesinkmetrics "github.com/pingcap/tiflow/cdc/sink/metrics/tablesink"
 	"github.com/pingcap/tiflow/cdc/sink/tablesink"
-	"github.com/pingcap/tiflow/pkg/config"
 	pconfig "github.com/pingcap/tiflow/pkg/config"
 	cerror "github.com/pingcap/tiflow/pkg/errors"
 	"github.com/pingcap/tiflow/pkg/retry"
@@ -330,12 +329,6 @@ func (m *SinkManager) Run(ctx context.Context, warnings ...chan<- error) (err er
 	}
 }
 
-func (m *SinkManager) needsStuckCheck() bool {
-	m.sinkFactory.Lock()
-	defer m.sinkFactory.Unlock()
-	return m.sinkFactory.f != nil && m.sinkFactory.f.Category() == factory.CategoryMQ
-}
-
 func (m *SinkManager) initSinkFactory() (chan error, uint64) {
 	m.sinkFactory.Lock()
 	defer m.sinkFactory.Unlock()
@@ -403,19 +396,6 @@ func (m *SinkManager) clearSinkFactory() {
 	}
 }
 
-func (m *SinkManager) putSinkFactoryError(err error, version uint64) (success bool) {
-	m.sinkFactory.Lock()
-	defer m.sinkFactory.Unlock()
-	if version == m.sinkFactory.version {
-		select {
-		case m.sinkFactory.errors <- err:
-		default:
-		}
-		return true
-	}
-	return false
-}
-
 func (m *SinkManager) startSinkWorkers(ctx context.Context, eg *errgroup.Group, splitTxn bool) {
 	for i := 0; i < sinkWorkerNum; i++ {
 		w := newSinkWorker(m.changefeedID, m.sourceManager,
@@ -1023,25 +1003,6 @@ func (m *SinkManager) GetTableStats(span tablepb.Span) TableStats {
 	m.sinkMemQuota.Release(span, checkpointTs)
 	m.redoMemQuota.Release(span, checkpointTs)
 
-	advanceTimeoutInSec := util.GetOrZero(m.config.Sink.AdvanceTimeoutInSec)
-	if advanceTimeoutInSec <= 0 {
-		advanceTimeoutInSec = config.DefaultAdvanceTimeoutInSec
-	}
-	stuckCheck := time.Duration(advanceTimeoutInSec) * time.Second
-
-	if m.needsStuckCheck() {
-		isStuck, sinkVersion := tableSink.sinkMaybeStuck(stuckCheck)
-		if isStuck && m.putSinkFactoryError(errors.New("table sink stuck"), sinkVersion) {
-			log.Warn("Table checkpoint is stuck too long, will restart the sink backend",
-				zap.String("namespace", m.changefeedID.Namespace),
-				zap.String("changefeed", m.changefeedID.ID),
-				zap.Stringer("span", &span),
-				zap.Any("checkpointTs", checkpointTs),
-				zap.Float64("stuckCheck", stuckCheck.Seconds()),
-				zap.Uint64("factoryVersion", sinkVersion))
-		}
-	}
-
 	var resolvedTs model.Ts
 	// If redo log is enabled, we have to use redo log's resolved ts to calculate processor's min resolved ts.
 	if m.redoDMLMgr != nil {

cdc/processor/sinkmanager/manager_test.go

@@ -378,21 +378,6 @@ func TestSinkManagerRunWithErrors(t *testing.T) {
 	}
 }
 
-func TestSinkManagerNeedsStuckCheck(t *testing.T) {
-	t.Parallel()
-
-	ctx, cancel := context.WithCancel(context.Background())
-	errCh := make(chan error, 16)
-	changefeedInfo := getChangefeedInfo()
-	manager, _, _ := CreateManagerWithMemEngine(t, ctx, model.DefaultChangeFeedID("1"), changefeedInfo, errCh)
-	defer func() {
-		cancel()
-		manager.Close()
-	}()
-
-	require.False(t, manager.needsStuckCheck())
-}
-
 func TestSinkManagerRestartTableSinks(t *testing.T) {
 	failpoint.Enable("github.com/pingcap/tiflow/cdc/processor/sinkmanager/SinkWorkerTaskHandlePause", "return")
 	defer failpoint.Disable("github.com/pingcap/tiflow/cdc/processor/sinkmanager/SinkWorkerTaskHandlePause")

cdc/processor/sinkmanager/table_sink_wrapper.go

@@ -453,25 +453,6 @@ func (t *tableSinkWrapper) cleanRangeEventCounts(upperBound sorter.Position, min
 	return shouldClean
 }
 
-func (t *tableSinkWrapper) sinkMaybeStuck(stuckCheck time.Duration) (bool, uint64) {
-	t.getCheckpointTs()
-
-	t.tableSink.RLock()
-	defer t.tableSink.RUnlock()
-	t.tableSink.innerMu.Lock()
-	defer t.tableSink.innerMu.Unlock()
-
-	// What these conditions mean:
-	// 1. the table sink has been associated with a valid sink;
-	// 2. its checkpoint hasn't been advanced for a while;
-	version := t.tableSink.version
-	advanced := t.tableSink.advanced
-	if version > 0 && time.Since(advanced) > stuckCheck {
-		return true, version
-	}
-	return false, uint64(0)
-}
-
 func handleRowChangedEvents(
 	changefeed model.ChangeFeedID, span tablepb.Span,
 	events ...*model.PolymorphicEvent,

cdc/processor/sinkmanager/table_sink_wrapper_test.go

@@ -18,7 +18,6 @@ import (
 	"math"
 	"sync"
 	"testing"
-	"time"
 
 	"github.com/pingcap/tiflow/cdc/model"
 	"github.com/pingcap/tiflow/cdc/processor/tablepb"
@@ -29,7 +28,6 @@ import (
 	"github.com/pingcap/tiflow/pkg/spanz"
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/stretchr/testify/require"
-	"github.com/tikv/client-go/v2/oracle"
 )
 
 type mockSink struct {
@@ -315,63 +313,3 @@ func TestTableSinkWrapperSinkVersion(t *testing.T) {
 	require.Nil(t, wrapper.tableSink.s)
 	require.Equal(t, wrapper.tableSink.version, uint64(0))
 }
-
-func TestTableSinkWrapperSinkInner(t *testing.T) {
-	t.Parallel()
-
-	innerTableSink := tablesink.New[*model.RowChangedEvent](
-		model.ChangeFeedID{}, tablepb.Span{}, model.Ts(0),
-		newMockSink(), &dmlsink.RowChangeEventAppender{},
-		pdutil.NewClock4Test(),
-		prometheus.NewCounter(prometheus.CounterOpts{}),
-		prometheus.NewHistogram(prometheus.HistogramOpts{}),
-	)
-	version := new(uint64)
-
-	wrapper := newTableSinkWrapper(
-		model.DefaultChangeFeedID("1"),
-		spanz.TableIDToComparableSpan(1),
-		func() (tablesink.TableSink, uint64) {
-			*version += 1
-			return innerTableSink, *version
-		},
-		tablepb.TableStatePrepared,
-		oracle.GoTimeToTS(time.Now()),
-		oracle.GoTimeToTS(time.Now().Add(10000*time.Second)),
-		func(_ context.Context) (model.Ts, error) { return math.MaxUint64, nil },
-	)
-
-	require.True(t, wrapper.initTableSink())
-
-	wrapper.closeAndClearTableSink()
-
-	// Shouldn't be stuck because version is 0.
-	require.Equal(t, wrapper.tableSink.version, uint64(0))
-	isStuck, _ := wrapper.sinkMaybeStuck(100 * time.Millisecond)
-	require.False(t, isStuck)
-
-	// Shouldn't be stuck because tableSink.advanced is just updated.
-	require.True(t, wrapper.initTableSink())
-	isStuck, _ = wrapper.sinkMaybeStuck(100 * time.Millisecond)
-	require.False(t, isStuck)
-
-	// Shouldn't be stuck because upperbound hasn't been advanced.
-	time.Sleep(200 * time.Millisecond)
-	isStuck, _ = wrapper.sinkMaybeStuck(100 * time.Millisecond)
-	require.False(t, isStuck)
-
-	// Shouldn't be stuck because `getCheckpointTs` will update tableSink.advanced.
-	nowTs := oracle.GoTimeToTS(time.Now())
-	wrapper.updateReceivedSorterResolvedTs(nowTs)
-	wrapper.barrierTs.Store(nowTs)
-	isStuck, _ = wrapper.sinkMaybeStuck(100 * time.Millisecond)
-	require.False(t, isStuck)
-
-	time.Sleep(200 * time.Millisecond)
-	nowTs = oracle.GoTimeToTS(time.Now())
-	wrapper.updateReceivedSorterResolvedTs(nowTs)
-	wrapper.barrierTs.Store(nowTs)
-	wrapper.updateResolvedTs(model.NewResolvedTs(nowTs))
-	isStuck, _ = wrapper.sinkMaybeStuck(100 * time.Millisecond)
-	require.True(t, isStuck)
-}

pkg/config/sink.go

@@ -178,6 +178,7 @@ type SinkConfig struct {
 
 	// AdvanceTimeoutInSec is a duration in second. If a table sink progress hasn't been
 	// advanced for this given duration, the sink will be canceled and re-established.
+	// Deprecated from v8.1.1.
 	AdvanceTimeoutInSec *uint `toml:"advance-timeout-in-sec" json:"advance-timeout-in-sec,omitempty"`
 
 	// Simple Protocol only config, use to control the behavior of sending bootstrap message.

pkg/sink/kafka/sarama_factory.go

@@ -15,12 +15,15 @@
 
 import (
 	"context"
+	"time"
 
 	"github.com/IBM/sarama"
+	"github.com/pingcap/log"
 	"github.com/pingcap/tiflow/cdc/model"
 	"github.com/pingcap/tiflow/pkg/errors"
 	"github.com/pingcap/tiflow/pkg/util"
 	"github.com/rcrowley/go-metrics"
+	"go.uber.org/zap"
 )
 
 type saramaFactory struct {
@@ -43,17 +46,32 @@
 }
 
 func (f *saramaFactory) AdminClient(ctx context.Context) (ClusterAdminClient, error) {
+	start := time.Now()
 	config, err := NewSaramaConfig(ctx, f.option)
+	duration := time.Since(start).Seconds()
+	if duration > 2 {
+		log.Warn("new sarama config cost too much time", zap.Any("duration", duration), zap.Stringer("changefeedID", f.changefeedID))
+	}
 	if err != nil {
 		return nil, err
 	}
 
+	start = time.Now()
 	client, err := sarama.NewClient(f.option.BrokerEndpoints, config)
+	duration = time.Since(start).Seconds()
+	if duration > 2 {
+		log.Warn("new sarama client cost too much time", zap.Any("duration", duration), zap.Stringer("changefeedID", f.changefeedID))
+	}
 	if err != nil {
 		return nil, errors.Trace(err)
 	}
 
+	start = time.Now()
 	admin, err := sarama.NewClusterAdminFromClient(client)
+	duration = time.Since(start).Seconds()
+	if duration > 2 {
+		log.Warn("new sarama cluster admin cost too much time", zap.Any("duration", duration), zap.Stringer("changefeedID", f.changefeedID))
+	}
 	if err != nil {
 		return nil, errors.Trace(err)
 	}

tests/integration_tests/run_group.sh

@@ -11,7 +11,7 @@ group_num=${group#G}
 # Other tests that only support mysql: batch_update_to_no_batch ddl_reentrant
 # changefeed_fast_fail changefeed_resume_with_checkpoint_ts sequence
 # multi_cdc_cluster capture_suicide_while_balance_table
-mysql_only="bdr_mode capture_suicide_while_balance_table syncpoint syncpoint_check_ts hang_sink_suicide server_config_compatibility changefeed_dup_error_restart"
+mysql_only="bdr_mode capture_suicide_while_balance_table syncpoint syncpoint_check_ts server_config_compatibility changefeed_dup_error_restart"
 mysql_only_http="http_api http_api_tls api_v2 http_api_tls_with_user_auth cli_tls_with_auth"
 mysql_only_consistent_replicate="consistent_replicate_ddl consistent_replicate_gbk consistent_replicate_nfs consistent_replicate_storage_file consistent_replicate_storage_file_large_value consistent_replicate_storage_s3 consistent_partition_table"