sorter(ticdc): fix dml order

cdc/model/mounter.go

@@ -131,6 +131,10 @@ func ComparePolymorphicEvents(i, j *PolymorphicEvent) bool {
 		if i.RawKV.OpType == OpTypeDelete && j.RawKV.OpType != OpTypeDelete {
 			return true
 		}
+		// update DML
+		if i.RawKV.OldValue != nil && j.RawKV.OldValue == nil {
+			return true
+		}
 	}
 	return i.CRTs < j.CRTs
 }

cdc/model/mounter_test.go

@@ -98,3 +98,33 @@ func TestResolvedTsEqual(t *testing.T) {
 	t5 := ResolvedTs{Mode: BatchResolvedMode, Ts: 2, BatchID: 1}
 	require.False(t, t1.Equal(t5))
 }
+
+func TestComparePolymorphicEvents(t *testing.T) {
+	cases := []struct {
+		a *PolymorphicEvent
+		b *PolymorphicEvent
+	}{
+		{
+			a: NewPolymorphicEvent(&RawKVEntry{
+				OpType: OpTypeDelete,
+			}),
+			b: NewPolymorphicEvent(&RawKVEntry{
+				OpType: OpTypePut,
+			}),
+		},
+		{
+			a: NewPolymorphicEvent(&RawKVEntry{
+				OpType:   OpTypePut,
+				OldValue: []byte{0},
+				Value:    []byte{0},
+			}),
+			b: NewPolymorphicEvent(&RawKVEntry{
+				OpType: OpTypePut,
+				Value:  []byte{0},
+			}),
+		},
+	}
+	for _, item := range cases {
+		require.True(t, ComparePolymorphicEvents(item.a, item.b))
+	}
+}

cdc/processor/sourcemanager/engine/pebble/encoding/key.go

@@ -21,6 +21,12 @@ import (
 	"go.uber.org/zap"
 )
 
+const (
+	typeDelete = iota + 1
+	typeUpdate
+	typeInsert
+)
+
 // DecodeKey decodes a key to uniqueID, tableID, startTs, CRTs.
 func DecodeKey(key []byte) (uniqueID uint32, tableID uint64, startTs, CRTs uint64) {
 	// uniqueID, tableID, CRTs, startTs, Key, Put/Delete
@@ -75,7 +81,7 @@ func EncodeTsKey(uniqueID uint32, tableID uint64, CRTs uint64, startTs ...uint64
 }
 
 // EncodeKey encodes a key according to event.
-// Format: uniqueID, tableID, CRTs, startTs, Put/Delete, Key.
+// Format: uniqueID, tableID, CRTs, startTs, delete/update/insert, Key.
 func EncodeKey(uniqueID uint32, tableID uint64, event *model.PolymorphicEvent) []byte {
 	if event.RawKV == nil {
 		log.Panic("rawkv must not be nil", zap.Any("event", event))
@@ -96,9 +102,19 @@ func EncodeKey(uniqueID uint32, tableID uint64, event *model.PolymorphicEvent) [
 	// startTs
 	binary.BigEndian.PutUint64(uint64Buf[:], event.StartTs)
 	buf = append(buf, uint64Buf[:]...)
-	// Let Delete < Put
-	binary.BigEndian.PutUint16(uint64Buf[:], ^uint16(event.RawKV.OpType))
+	// Let Delete < Update < Insert
+	binary.BigEndian.PutUint16(uint64Buf[:], getDMLOrder(event.RawKV))
 	buf = append(buf, uint64Buf[:2]...)
 	// key
 	return append(buf, event.RawKV.Key...)
 }
+
+// getDMLOrder returns the order of the dml types: delete<update<insert
+func getDMLOrder(rowKV *model.RawKVEntry) uint16 {
+	if rowKV.OpType == model.OpTypeDelete {
+		return typeDelete
+	} else if rowKV.OldValue != nil {
+		return typeUpdate
+	}
+	return typeInsert
+}

cdc/processor/sourcemanager/engine/pebble/encoding/key_test.go

@@ -133,6 +133,27 @@ func TestEncodeKey(t *testing.T) {
 			CRTs:    3,
 		}),
 	)
+
+	// update < insert
+	mustLess(
+		0, 0,
+		1, 1,
+		model.NewPolymorphicEvent(&model.RawKVEntry{
+			OpType:   model.OpTypePut,
+			Key:      []byte{2},
+			StartTs:  1,
+			CRTs:     1,
+			Value:    []byte{2},
+			OldValue: []byte{2},
+		}),
+		model.NewPolymorphicEvent(&model.RawKVEntry{
+			OpType:  model.OpTypePut,
+			Key:     []byte{2},
+			StartTs: 1,
+			CRTs:    1,
+			Value:   []byte{2},
+		}),
+	)
 }
 
 func TestEncodeTsKey(t *testing.T) {