fix filter skip checkpoint problem

drainer/syncer.go

@@ -21,6 +21,7 @@ import (
 
 	"github.com/pingcap/tidb-binlog/drainer/loopbacksync"
 	"github.com/pingcap/tidb-binlog/pkg/loader"
+	"github.com/pingcap/tidb-binlog/pkg/util"
 
 	"github.com/pingcap/errors"
 	"github.com/pingcap/log"
@@ -290,14 +291,27 @@ func (s *Syncer) run() error {
 		return err
 	}
 
-	var lastDDLSchemaVersion int64
-	var b *binlogItem
+	var (
+		lastDDLSchemaVersion int64
+		b                    *binlogItem
+		fakeBinlog           *pb.Binlog
+		pushFakeBinlog       chan<- *pb.Binlog
+		lastAddCommitTS      int64
+		lastFakeCommitTime   time.Time
+	)
+	dsyncError := s.dsyncer.Error()
 
-	var fakeBinlog *pb.Binlog
-	var pushFakeBinlog chan<- *pb.Binlog
+	appendFakeBinlogIfNeeded := func(binlog *pb.Binlog, commitTS int64) {
+		if fakeCommitTime := oracle.GetTimeFromTS(uint64(commitTS)); fakeCommitTime.Sub(lastFakeCommitTime) > 3*time.Second {
+			lastFakeCommitTime = fakeCommitTime
+			if binlog == nil {
+				binlog = util.GenFakeBinlog(commitTS)
+			}
+			fakeBinlogs = append(fakeBinlogs, binlog)
+			fakeBinlogPreAddTS = append(fakeBinlogPreAddTS, lastAddCommitTS)
+		}
+	}
 
-	var lastAddComitTS int64
-	dsyncError := s.dsyncer.Error()
 ForLoop:
 	for {
 		// check if we can safely push a fake binlog
@@ -335,8 +349,7 @@ ForLoop:
 		}
 
 		if startTS == commitTS {
-			fakeBinlogs = append(fakeBinlogs, binlog)
-			fakeBinlogPreAddTS = append(fakeBinlogPreAddTS, lastAddComitTS)
+			appendFakeBinlogIfNeeded(binlog, commitTS)
 		} else if jobID == 0 {
 			preWriteValue := binlog.GetPrewriteValue()
 			preWrite := &pb.PrewriteValue{}
@@ -363,39 +376,45 @@ ForLoop:
 				break ForLoop
 			}
 
-			var isFilterTransaction = false
-			var err1 error
+			var (
+				isFilterTransaction = false
+				ignore              = false
+				err1                error
+			)
 			if s.loopbackSync != nil && s.loopbackSync.LoopbackControl {
 				isFilterTransaction, err1 = loopBackStatus(binlog, preWrite, s.schema, s.loopbackSync)
 				if err1 != nil {
 					err = errors.Annotate(err1, "analyze transaction failed")
 					break ForLoop
 				}
 			}
-
-			var ignore bool
-			ignore, err = filterTable(preWrite, s.filter, s.schema)
-			if err != nil {
-				err = errors.Annotate(err, "filterTable failed")
-				break ForLoop
+			if !isFilterTransaction {
+				ignore, err = filterTable(preWrite, s.filter, s.schema)
+				if err != nil {
+					err = errors.Annotate(err, "filterTable failed")
+					break ForLoop
+				}
 			}
 
 			if !ignore && !isFilterTransaction {
 				s.addDMLEventMetrics(preWrite.GetMutations())
 				beginTime := time.Now()
-				lastAddComitTS = binlog.GetCommitTs()
+				lastAddCommitTS = binlog.GetCommitTs()
 				err = s.dsyncer.Sync(&dsync.Item{Binlog: binlog, PrewriteValue: preWrite, SchemaVersion: preWrite.SchemaVersion})
 				if err != nil {
 					err = errors.Annotatef(err, "failed to add item")
 					break ForLoop
 				}
 				executeHistogram.Observe(time.Since(beginTime).Seconds())
+			} else {
+				appendFakeBinlogIfNeeded(nil, commitTS)
 			}
 		} else if jobID > 0 {
 			log.Debug("get ddl binlog job", zap.Stringer("job", b.job))
 
 			if skipUnsupportedDDLJob(b.job) {
 				log.Info("skip unsupported DDL job", zap.Stringer("job", b.job))
+				appendFakeBinlogIfNeeded(nil, commitTS)
 				continue
 			}
 
@@ -413,6 +432,7 @@ ForLoop:
 
 			if b.job.SchemaState == model.StateDeleteOnly && b.job.Type == model.ActionDropColumn {
 				log.Info("Syncer skips DeleteOnly DDL", zap.Stringer("job", b.job), zap.Int64("ts", b.GetCommitTs()))
+				appendFakeBinlogIfNeeded(nil, commitTS)
 				continue
 			}
 
@@ -427,6 +447,7 @@ ForLoop:
 			if s.filter.SkipSchemaAndTable(schema, table) {
 				log.Info("skip ddl by filter", zap.String("schema", schema), zap.String("table", table),
 					zap.String("sql", sql), zap.Int64("commit ts", commitTS))
+				appendFakeBinlogIfNeeded(nil, commitTS)
 				continue
 			}
 
@@ -439,14 +460,15 @@ ForLoop:
 				if s.cfg.DestDBType == "tidb" || s.cfg.DestDBType == "mysql" {
 					shouldSkip = true
 				} else {
+					appendFakeBinlogIfNeeded(nil, commitTS)
 					continue
 				}
 			}
 
 			// Add ddl item to downstream.
 			s.addDDLCount()
 			beginTime := time.Now()
-			lastAddComitTS = binlog.GetCommitTs()
+			lastAddCommitTS = binlog.GetCommitTs()
 
 			log.Info("add ddl item to syncer, you can add this commit ts to `ignore-txn-commit-ts` to skip this ddl if needed",
 				zap.String("sql", sql), zap.Int64("commit ts", binlog.CommitTs))

pkg/util/ts.go

@@ -19,6 +19,7 @@ import (
 	"github.com/pingcap/errors"
 	"github.com/pingcap/log"
 	"github.com/pingcap/tidb/store/tikv/oracle"
+	"github.com/pingcap/tipb/go-binlog"
 	pd "github.com/tikv/pd/client"
 	"go.uber.org/zap"
 	"golang.org/x/net/context"
@@ -55,6 +56,15 @@ func GetTSO(pdCli pd.Client) (int64, error) {
 	return ts, nil
 }
 
+// GenFakeBinlog generates a fake binlog from given tso
+func GenFakeBinlog(ts int64) *binlog.Binlog {
+	return &binlog.Binlog{
+		StartTs:  ts,
+		Tp:       binlog.BinlogType_Rollback,
+		CommitTs: ts,
+	}
+}
+
 // TSOToRoughTime translates tso to rough time that used to display
 func TSOToRoughTime(ts int64) time.Time {
 	t := time.Unix(ts>>18/1000, 0)

pump/server.go

@@ -469,13 +469,7 @@ func (s *Server) genFakeBinlog() (*pb.Binlog, error) {
 		return nil, errors.Trace(err)
 	}
 
-	bl := &binlog.Binlog{
-		StartTs:  ts,
-		Tp:       binlog.BinlogType_Rollback,
-		CommitTs: ts,
-	}
-
-	return bl, nil
+	return util.GenFakeBinlog(ts), nil
 }
 
 func (s *Server) writeFakeBinlog() (*pb.Binlog, error) {