Add segment reference count and handles change info in ShardCluster (#16620)

Resolves #16619
Add reference count for each search/query request
For SegmentChangeInfo
- Wait all segments in OnlineList to be loaded
- Add handoff event into pending list
- Wait all segments in OfflineList is not used (reference count = 0)

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>

internal/querynode/query_channel.go

@@ -20,10 +20,23 @@ type queryChannel struct {
 
 	streaming      *streaming
 	queryMsgStream msgstream.MsgStream
+	shardCluster   *ShardClusterService
 	asConsumeOnce  sync.Once
 	closeOnce      sync.Once
 }
 
+// NewQueryChannel create a query channel with provided shardCluster, query msgstream and collection id
+func NewQueryChannel(collectionID int64, scs *ShardClusterService, qms msgstream.MsgStream, streaming *streaming) *queryChannel {
+	return &queryChannel{
+		closeCh:      make(chan struct{}),
+		collectionID: collectionID,
+
+		streaming:      streaming,
+		queryMsgStream: qms,
+		shardCluster:   scs,
+	}
+}
+
 // AsConsumer do AsConsumer for query msgstream and seek if position is not nil
 func (qc *queryChannel) AsConsumer(channelName string, subName string, position *internalpb.MsgPosition) error {
 	var err error
@@ -101,7 +114,9 @@ func (qc *queryChannel) adjustByChangeInfo(msg *msgstream.SealedSegmentsChangeIn
 			}
 		}
 
-		// should handle segment change in shardCluster
+		// process change in shard cluster
+		qc.shardCluster.HandoffSegments(qc.collectionID, info)
+
 		// for OnlineSegments:
 		for _, segment := range info.OnlineSegments {
 			/*
@@ -124,12 +139,6 @@ func (qc *queryChannel) adjustByChangeInfo(msg *msgstream.SealedSegmentsChangeIn
 				},
 			})
 		}
-		/*
-			// for OfflineSegments:
-			for _, segment := range info.OfflineSegments {
-				// 1. update global sealed segments
-				q.globalSegmentManager.removeGlobalSealedSegmentInfo(segment.SegmentID)
-			}*/
 
 		log.Info("Successfully changed global sealed segment info ",
 			zap.Int64("collection ", qc.collectionID),

internal/querynode/query_channel_test.go

@@ -97,13 +97,7 @@ func TestQueryChannel_AsConsumer(t *testing.T) {
 		mqs := &mockQueryMsgStream{}
 		mqs.On("Close").Return()
 
-		qc := &queryChannel{
+		qc := NewQueryChannel(defaultCollectionID, nil, mqs, nil)
-			closeCh:      make(chan struct{}),
-			collectionID: defaultCollectionID,
-
-			streaming:      nil,
-			queryMsgStream: mqs,
-		}
 
 		mqs.On("AsConsumer", []string{defaultDMLChannel}, defaultSubName).Return()
 
@@ -122,13 +116,7 @@ func TestQueryChannel_AsConsumer(t *testing.T) {
 		mqs := &mockQueryMsgStream{}
 		mqs.On("Close").Return()
 
-		qc := &queryChannel{
+		qc := NewQueryChannel(defaultCollectionID, nil, mqs, nil)
-			closeCh:      make(chan struct{}),
-			collectionID: defaultCollectionID,
-
-			streaming:      nil,
-			queryMsgStream: mqs,
-		}
 
 		mqs.On("AsConsumer", []string{defaultDMLChannel}, defaultSubName).Return()
 
@@ -146,13 +134,8 @@ func TestQueryChannel_AsConsumer(t *testing.T) {
 		mqs := &mockQueryMsgStream{}
 		mqs.On("Close").Return()
 
-		qc := &queryChannel{
+		qc := NewQueryChannel(defaultCollectionID, nil, mqs, nil)
-			closeCh:      make(chan struct{}),
-			collectionID: defaultCollectionID,
 
-			streaming:      nil,
-			queryMsgStream: mqs,
-		}
 		msgID := make([]byte, 8)
 		rand.Read(msgID)
 		pos := &internalpb.MsgPosition{MsgID: msgID}

internal/querynode/query_shard_service.go

@@ -138,12 +138,7 @@ func (q *queryShardService) getQueryChannel(collectionID int64) *queryChannel {
 	qc, ok := q.queryChannels[collectionID]
 	if !ok {
 		queryStream, _ := q.factory.NewQueryMsgStream(q.ctx)
-		qc = &queryChannel{
+		qc = NewQueryChannel(collectionID, q.shardClusterService, queryStream, q.streaming)
-			closeCh:        make(chan struct{}),
-			collectionID:   collectionID,
-			queryMsgStream: queryStream,
-			streaming:      q.streaming,
-		}
 		q.queryChannels[collectionID] = qc
 	}
 

internal/querynode/shard_cluster.go

@@ -92,6 +92,7 @@ type shardSegmentInfo struct {
 	partitionID int64
 	nodeID      int64
 	state       segmentState
+	inUse       int32
 }
 
 // ShardNodeDetector provides method to detect node events
@@ -122,6 +123,10 @@ type ShardCluster struct {
 	mut         sync.RWMutex
 	nodes       map[int64]*shardNode                 // online nodes
 	segments    map[int64]*shardSegmentInfo          // shard segments
+	handoffs    map[int32]*querypb.SegmentChangeInfo // current pending handoff
+	lastToken   *atomic.Int32                        // last token used for segment change info
+	segmentCond *sync.Cond                           // segment state change condition
+	rcCond      *sync.Cond                           // segment rc change condition
 
 	closeOnce sync.Once
 	closeCh   chan struct{}
@@ -143,10 +148,17 @@ func NewShardCluster(collectionID int64, replicaID int64, vchannelName string,
 
 		nodes:     make(map[int64]*shardNode),
 		segments:  make(map[int64]*shardSegmentInfo),
+		handoffs:  make(map[int32]*querypb.SegmentChangeInfo),
+		lastToken: atomic.NewInt32(0),
 
 		closeCh: make(chan struct{}),
 	}
 
+	m := sync.Mutex{}
+	sc.segmentCond = sync.NewCond(&m)
+	m2 := sync.Mutex{}
+	sc.rcCond = sync.NewCond(&m2)
+
 	sc.init()
 
 	return sc
@@ -205,8 +217,15 @@ func (sc *ShardCluster) removeNode(evt nodeEvent) {
 
 // updateSegment apply segment change to shard cluster
 func (sc *ShardCluster) updateSegment(evt segmentEvent) {
-
 	log.Debug("ShardCluster update segment", zap.Int64("nodeID", evt.nodeID), zap.Int64("segmentID", evt.segmentID), zap.Int32("state", int32(evt.state)))
+
+	// notify handoff wait online if any
+	defer func() {
+		sc.segmentCond.L.Lock()
+		sc.segmentCond.Broadcast()
+		sc.segmentCond.L.Unlock()
+	}()
+
 	sc.mut.Lock()
 	defer sc.mut.Unlock()
 
@@ -255,6 +274,8 @@ func (sc *ShardCluster) transferSegment(old *shardSegmentInfo, evt segmentEvent)
 // removeSegment removes segment from cluster
 // should only applied in hand-off or load balance procedure
 func (sc *ShardCluster) removeSegment(evt segmentEvent) {
+	log.Debug("ShardCluster remove segment", zap.Int64("nodeID", evt.nodeID), zap.Int64("segmentID", evt.segmentID), zap.Int32("state", int32(evt.state)))
+
 	sc.mut.Lock()
 	defer sc.mut.Unlock()
 
@@ -269,7 +290,6 @@ func (sc *ShardCluster) removeSegment(evt segmentEvent) {
 		return
 	}
 
-	//TODO check handoff / load balance
 	delete(sc.segments, evt.segmentID)
 }
 
@@ -331,7 +351,6 @@ func (sc *ShardCluster) watchSegments(evtCh <-chan segmentEvent) {
 	for {
 		select {
 		case evt, ok := <-evtCh:
-			log.Debug("segment event", zap.Any("evt", evt))
 			if !ok {
 				log.Warn("ShardCluster segment channel closed", zap.Int64("collectionID", sc.collectionID), zap.Int64("replicaID", sc.replicaID))
 				return
@@ -381,20 +400,166 @@ func (sc *ShardCluster) getSegment(segmentID int64) (*shardSegmentInfo, bool) {
 }
 
 // segmentAllocations returns node to segments mappings.
+// calling this function also increases the reference count of related segments.
 func (sc *ShardCluster) segmentAllocations(partitionIDs []int64) map[int64][]int64 {
 	result := make(map[int64][]int64) // nodeID => segmentIDs
-	sc.mut.RLock()
+	sc.mut.Lock()
-	defer sc.mut.RUnlock()
+	defer sc.mut.Unlock()
 
 	for _, segment := range sc.segments {
 		if len(partitionIDs) > 0 && !inList(partitionIDs, segment.partitionID) {
 			continue
 		}
+		if sc.inHandoffOffline(segment.segmentID) {
+			log.Debug("segment ignore in pending offline list", zap.Int64("collectionID", sc.collectionID), zap.Int64("replicaID", sc.replicaID), zap.Int64("segmentID", segment.segmentID))
+			continue
+		}
+		// reference count ++
+		segment.inUse++
 		result[segment.nodeID] = append(result[segment.nodeID], segment.segmentID)
 	}
 	return result
 }
 
+// inHandoffOffline checks whether segment is pending handoff offline list
+// Note that sc.mut Lock is assumed to be hold outside of this function!
+func (sc *ShardCluster) inHandoffOffline(segmentID int64) bool {
+	for _, handoff := range sc.handoffs {
+		for _, offlineSegment := range handoff.OfflineSegments {
+			if segmentID == offlineSegment.GetSegmentID() {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// finishUsage decreases the inUse count of provided segments
+func (sc *ShardCluster) finishUsage(allocs map[int64][]int64) {
+	defer func() {
+		sc.rcCond.L.Lock()
+		sc.rcCond.Broadcast()
+		sc.rcCond.L.Unlock()
+	}()
+	sc.mut.Lock()
+	defer sc.mut.Unlock()
+	for _, segments := range allocs {
+		for _, segmentID := range segments {
+			segment, ok := sc.segments[segmentID]
+			if !ok || segment == nil {
+				// this shall not happen, since removing segment without decreasing rc to zero is illegal
+				log.Error("finishUsage with non-existing segment", zap.Int64("collectionID", sc.collectionID), zap.Int64("replicaID", sc.replicaID), zap.String("vchannel", sc.vchannelName), zap.Int64("segmentID", segmentID))
+				continue
+			}
+			// decrease the reference count
+			segment.inUse--
+		}
+	}
+}
+
+// HandoffSegments processes the handoff/load balance segments update procedure.
+func (sc *ShardCluster) HandoffSegments(info *querypb.SegmentChangeInfo) error {
+	// wait for all OnlineSegment is loaded
+	onlineSegments := make([]int64, 0, len(info.OnlineSegments))
+	for _, seg := range info.OnlineSegments {
+		// filter out segments not maintained in this cluster
+		if seg.GetCollectionID() != sc.collectionID || seg.GetDmChannel() != sc.vchannelName {
+			continue
+		}
+		onlineSegments = append(onlineSegments, seg.GetSegmentID())
+	}
+	sc.waitSegmentsOnline(onlineSegments)
+
+	// add segmentChangeInfo to pending list
+	token := sc.appendHandoff(info)
+
+	// wait for all OfflineSegments is not in use
+	offlineSegments := make([]int64, 0, len(info.OfflineSegments))
+	for _, seg := range info.OfflineSegments {
+		offlineSegments = append(offlineSegments, seg.GetSegmentID())
+	}
+	sc.waitSegmentsNotInUse(offlineSegments)
+	// remove offline segments record
+	for _, seg := range info.OfflineSegments {
+		// filter out segments not maintained in this cluster
+		if seg.GetCollectionID() != sc.collectionID || seg.GetDmChannel() != sc.vchannelName {
+			continue
+		}
+		sc.removeSegment(segmentEvent{segmentID: seg.GetSegmentID(), nodeID: seg.GetNodeID()})
+	}
+
+	// finish handoff and remove it from pending list
+	sc.finishHandoff(token)
+
+	return nil
+}
+
+// appendHandoff adds the change info into pending list and returns the token.
+func (sc *ShardCluster) appendHandoff(info *querypb.SegmentChangeInfo) int32 {
+	sc.mut.Lock()
+	defer sc.mut.Unlock()
+
+	token := sc.lastToken.Add(1)
+	sc.handoffs[token] = info
+	return token
+}
+
+// finishHandoff removes the handoff related to the token.
+func (sc *ShardCluster) finishHandoff(token int32) {
+	sc.mut.Lock()
+	defer sc.mut.Unlock()
+
+	delete(sc.handoffs, token)
+}
+
+// waitSegmentsOnline waits until all provided segments is loaded.
+func (sc *ShardCluster) waitSegmentsOnline(segments []int64) {
+	sc.segmentCond.L.Lock()
+	for !sc.segmentsOnline(segments) {
+		sc.segmentCond.Wait()
+	}
+	sc.segmentCond.L.Unlock()
+}
+
+// waitSegmentsNotInUse waits until all provided segments is not in use.
+func (sc *ShardCluster) waitSegmentsNotInUse(segments []int64) {
+	sc.rcCond.L.Lock()
+	for sc.segmentsInUse(segments) {
+		sc.rcCond.Wait()
+	}
+	sc.rcCond.L.Unlock()
+}
+
+// checkOnline checks whether all segment ids provided in online state.
+func (sc *ShardCluster) segmentsOnline(segmentIDs []int64) bool {
+	sc.mut.RLock()
+	defer sc.mut.RUnlock()
+	for _, segID := range segmentIDs {
+		segment, ok := sc.segments[segID]
+		if !ok || segment.state != segmentStateLoaded {
+			return false
+		}
+	}
+	return true
+}
+
+// segmentsInUse checks whether all segment ids provided still in use.
+func (sc *ShardCluster) segmentsInUse(segmentIDs []int64) bool {
+	sc.mut.RLock()
+	defer sc.mut.RUnlock()
+	for _, segID := range segmentIDs {
+		segment, ok := sc.segments[segID]
+		if !ok {
+			// ignore missing segments, since they might be in streaming
+			continue
+		}
+		if segment.inUse > 0 {
+			return true
+		}
+	}
+	return false
+}
+
 // Search preforms search operation on shard cluster.
 func (sc *ShardCluster) Search(ctx context.Context, req *querypb.SearchRequest) ([]*internalpb.SearchResults, error) {
 	if sc.state.Load() != int32(available) {
@@ -405,16 +570,14 @@ func (sc *ShardCluster) Search(ctx context.Context, req *querypb.SearchRequest)
 		return nil, fmt.Errorf("ShardCluster for %s does not match to request channel :%s", sc.vchannelName, req.GetDmlChannel())
 	}
 
-	//req.GetReq().GetPartitionIDs()
+	// get node allocation and maintains the inUse reference count
-
-	// get node allocation
 	segAllocs := sc.segmentAllocations(req.GetReq().GetPartitionIDs())
+	defer sc.finishUsage(segAllocs)
 
 	log.Debug("cluster segment distribution", zap.Int("len", len(segAllocs)))
 	for nodeID, segmentIDs := range segAllocs {
 		log.Debug("segments distribution", zap.Int64("nodeID", nodeID), zap.Int64s("segments", segmentIDs))
 	}
-	// TODO dispatch to local queryShardService query dml channel growing segments
 
 	// concurrent visiting nodes
 	var wg sync.WaitGroup
@@ -430,7 +593,7 @@ func (sc *ShardCluster) Search(ctx context.Context, req *querypb.SearchRequest)
 		nodeReq.SegmentIDs = segments
 		node, ok := sc.getNode(nodeID)
 		if !ok { // meta dismatch, report error
-			return nil, fmt.Errorf("SharcCluster for %s replicaID %d is no available", sc.vchannelName, sc.replicaID)
+			return nil, fmt.Errorf("ShardCluster for %s replicaID %d is no available", sc.vchannelName, sc.replicaID)
 		}
 		wg.Add(1)
 		go func() {
@@ -466,10 +629,9 @@ func (sc *ShardCluster) Query(ctx context.Context, req *querypb.QueryRequest) ([
 		return nil, fmt.Errorf("ShardCluster for %s does not match to request channel :%s", sc.vchannelName, req.GetDmlChannel())
 	}
 
-	// get node allocation
+	// get node allocation and maintains the inUse reference count
 	segAllocs := sc.segmentAllocations(req.GetReq().GetPartitionIDs())
-
+	defer sc.finishUsage(segAllocs)
-	// TODO dispatch to local queryShardService query dml channel growing segments
 
 	// concurrent visiting nodes
 	var wg sync.WaitGroup

internal/querynode/shard_cluster_service.go

@@ -8,6 +8,7 @@ import (
 	"sync"
 
 	grpcquerynodeclient "github.com/milvus-io/milvus/internal/distributed/querynode/client"
+	"github.com/milvus-io/milvus/internal/proto/querypb"
 	"github.com/milvus-io/milvus/internal/util"
 	"github.com/milvus-io/milvus/internal/util/sessionutil"
 	"github.com/milvus-io/milvus/internal/util/typeutil"
@@ -107,3 +108,20 @@ func (s *ShardClusterService) releaseCollection(collectionID int64) {
 		return true
 	})
 }
+
+// HandoffSegments dispatch segmentChangeInfo to related shardClusters
+func (s *ShardClusterService) HandoffSegments(collectionID int64, info *querypb.SegmentChangeInfo) {
+	var wg sync.WaitGroup
+	s.clusters.Range(func(k, v interface{}) bool {
+		cs := v.(*ShardCluster)
+		if cs.collectionID == collectionID {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				cs.HandoffSegments(info)
+			}()
+		}
+		return true
+	})
+	wg.Wait()
+}

internal/querynode/shard_cluster_service_test.go

@@ -4,8 +4,10 @@ import (
 	"context"
 	"testing"
 
+	"github.com/milvus-io/milvus/internal/proto/querypb"
 	"github.com/milvus-io/milvus/internal/util/sessionutil"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 	"go.etcd.io/etcd/server/v3/etcdserver/api/v3client"
 )
 
@@ -32,3 +34,19 @@ func TestShardClusterService(t *testing.T) {
 	err = clusterService.releaseShardCluster("non-exist-channel")
 	assert.Error(t, err)
 }
+
+func TestShardClusterService_HandoffSegments(t *testing.T) {
+	qn, err := genSimpleQueryNode(context.Background())
+	require.NoError(t, err)
+
+	client := v3client.New(embedetcdServer.Server)
+	defer client.Close()
+	session := sessionutil.NewSession(context.Background(), "/by-dev/sessions/unittest/querynode/", client)
+	clusterService := newShardClusterService(client, session, qn)
+
+	clusterService.addShardCluster(defaultCollectionID, defaultReplicaID, defaultDMLChannel)
+	//TODO change shardCluster to interface to mock test behavior
+	assert.NotPanics(t, func() {
+		clusterService.HandoffSegments(defaultCollectionID, &querypb.SegmentChangeInfo{})
+	})
+}

internal/querynode/shard_cluster_test.go

@@ -1023,3 +1023,372 @@ func TestShardCluster_Query(t *testing.T) {
 	})
 
 }
+
+func TestShardCluster_ReferenceCount(t *testing.T) {
+	collectionID := int64(1)
+	vchannelName := "dml_1_1_v0"
+	replicaID := int64(0)
+	//	ctx := context.Background()
+
+	t.Run("normal alloc & finish", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		allocs := sc.segmentAllocations(nil)
+
+		sc.mut.RLock()
+		for _, segment := range sc.segments {
+			assert.Greater(t, segment.inUse, int32(0))
+		}
+		sc.mut.RUnlock()
+
+		assert.True(t, sc.segmentsInUse([]int64{1, 2}))
+		assert.True(t, sc.segmentsInUse([]int64{1, 2, -1}))
+
+		sc.finishUsage(allocs)
+		sc.mut.RLock()
+		for _, segment := range sc.segments {
+			assert.EqualValues(t, segment.inUse, 0)
+		}
+		sc.mut.RUnlock()
+
+		assert.False(t, sc.segmentsInUse([]int64{1, 2}))
+		assert.False(t, sc.segmentsInUse([]int64{1, 2, -1}))
+	})
+
+	t.Run("alloc & finish with modified alloc", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		allocs := sc.segmentAllocations(nil)
+
+		sc.mut.RLock()
+		for _, segment := range sc.segments {
+			assert.Greater(t, segment.inUse, int32(0))
+		}
+		sc.mut.RUnlock()
+
+		for node, segments := range allocs {
+			segments = append(segments, -1) // add non-exist segment
+			// shall be ignored in finishUsage
+			allocs[node] = segments
+		}
+
+		assert.NotPanics(t, func() {
+			sc.finishUsage(allocs)
+		})
+		sc.mut.RLock()
+		for _, segment := range sc.segments {
+			assert.EqualValues(t, segment.inUse, 0)
+		}
+		sc.mut.RUnlock()
+	})
+
+	t.Run("wait segments online", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		evtCh := make(chan segmentEvent, 10)
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+				evtCh:        evtCh,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		assert.True(t, sc.segmentsOnline([]int64{1, 2}))
+		assert.False(t, sc.segmentsOnline([]int64{1, 2, 3}))
+
+		sig := make(chan struct{})
+		go func() {
+			sc.waitSegmentsOnline([]int64{1, 2, 3})
+			close(sig)
+		}()
+
+		evtCh <- segmentEvent{
+			eventType: segmentAdd,
+			segmentID: 3,
+			nodeID:    1,
+			state:     segmentStateLoaded,
+		}
+
+		<-sig
+		assert.True(t, sc.segmentsOnline([]int64{1, 2, 3}))
+	})
+}
+
+func TestShardCluster_HandoffSegments(t *testing.T) {
+	collectionID := int64(1)
+	otherCollectionID := int64(2)
+	vchannelName := "dml_1_1_v0"
+	otherVchannelName := "dml_1_2_v0"
+	replicaID := int64(0)
+
+	t.Run("handoff without using segments", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		sc.HandoffSegments(&querypb.SegmentChangeInfo{
+			OnlineSegments: []*querypb.SegmentInfo{
+				{SegmentID: 2, NodeID: 2, CollectionID: collectionID, DmChannel: vchannelName},
+			},
+			OfflineSegments: []*querypb.SegmentInfo{
+				{SegmentID: 1, NodeID: 1, CollectionID: collectionID, DmChannel: vchannelName},
+			},
+		})
+
+		sc.mut.RLock()
+		_, has := sc.segments[1]
+		sc.mut.RUnlock()
+
+		assert.False(t, has)
+	})
+	t.Run("handoff with growing segment(segment not recorded)", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		sc.HandoffSegments(&querypb.SegmentChangeInfo{
+			OnlineSegments: []*querypb.SegmentInfo{
+				{SegmentID: 2, NodeID: 2, CollectionID: collectionID, DmChannel: vchannelName},
+				{SegmentID: 4, NodeID: 2, CollectionID: otherCollectionID, DmChannel: otherVchannelName},
+			},
+			OfflineSegments: []*querypb.SegmentInfo{
+				{SegmentID: 3, NodeID: 1, CollectionID: collectionID, DmChannel: vchannelName},
+				{SegmentID: 5, NodeID: 2, CollectionID: otherCollectionID, DmChannel: otherVchannelName},
+			},
+		})
+
+		sc.mut.RLock()
+		_, has := sc.segments[3]
+		sc.mut.RUnlock()
+
+		assert.False(t, has)
+	})
+	t.Run("handoff wait online and usage", func(t *testing.T) {
+		nodeEvents := []nodeEvent{
+			{
+				nodeID:   1,
+				nodeAddr: "addr_1",
+			},
+			{
+				nodeID:   2,
+				nodeAddr: "addr_2",
+			},
+		}
+
+		segmentEvents := []segmentEvent{
+			{
+				segmentID: 1,
+				nodeID:    1,
+				state:     segmentStateLoaded,
+			},
+			{
+				segmentID: 2,
+				nodeID:    2,
+				state:     segmentStateLoaded,
+			},
+		}
+		evtCh := make(chan segmentEvent, 10)
+		sc := NewShardCluster(collectionID, replicaID, vchannelName,
+			&mockNodeDetector{
+				initNodes: nodeEvents,
+			}, &mockSegmentDetector{
+				initSegments: segmentEvents,
+				evtCh:        evtCh,
+			}, buildMockQueryNode)
+		defer sc.Close()
+
+		// add rc to all segments
+		allocs := sc.segmentAllocations(nil)
+
+		sig := make(chan struct{})
+		go func() {
+			sc.HandoffSegments(&querypb.SegmentChangeInfo{
+				OnlineSegments: []*querypb.SegmentInfo{
+					{SegmentID: 3, NodeID: 1, CollectionID: collectionID, DmChannel: vchannelName},
+				},
+				OfflineSegments: []*querypb.SegmentInfo{
+					{SegmentID: 1, NodeID: 1, CollectionID: collectionID, DmChannel: vchannelName},
+				},
+			})
+
+			close(sig)
+		}()
+
+		sc.mut.RLock()
+		// still waiting online
+		assert.Equal(t, 0, len(sc.handoffs))
+		sc.mut.RUnlock()
+
+		evtCh <- segmentEvent{
+			eventType: segmentAdd,
+			segmentID: 3,
+			nodeID:    1,
+			state:     segmentStateLoaded,
+		}
+
+		// wait for handoff appended into list
+		assert.Eventually(t, func() bool {
+			sc.mut.RLock()
+			defer sc.mut.RUnlock()
+			return len(sc.handoffs) > 0
+		}, time.Second, time.Millisecond*10)
+
+		tmpAllocs := sc.segmentAllocations(nil)
+		found := false
+		for _, segments := range tmpAllocs {
+			if inList(segments, int64(1)) {
+				found = true
+				break
+			}
+		}
+		// segment 1 shall not be allocated again!
+		assert.False(t, found)
+		sc.finishUsage(tmpAllocs)
+		// rc shall be 0 now
+		sc.finishUsage(allocs)
+
+		// wait handoff finished
+		<-sig
+
+		sc.mut.RLock()
+		_, has := sc.segments[1]
+		sc.mut.RUnlock()
+
+		assert.False(t, has)
+	})
+}