Add log for allocator

Signed-off-by: Nzhenshan.cao <zhenshan.cao@zilliz.com>

internal/allocator/allocator.go

@@ -2,10 +2,13 @@ package allocator
 
 import (
 	"context"
+	"errors"
+	"fmt"
 	"sync"
 	"time"
 
-	"errors"
+	"github.com/zilliztech/milvus-distributed/internal/log"
+	"go.uber.org/zap"
 )
 
 const (
@@ -106,11 +109,13 @@ type Allocator struct {
 	TChan         TickerChan
 	ForceSyncChan chan Request
 
-	SyncFunc    func() bool
+	SyncFunc    func() (bool, error)
 	ProcessFunc func(req Request) error
 
 	CheckSyncFunc func(timeout bool) bool
 	PickCanDoFunc func()
+	SyncErr       error
+	Role          string
 }
 
 func (ta *Allocator) Start() error {
@@ -183,7 +188,7 @@ func (ta *Allocator) pickCanDo() {
 func (ta *Allocator) sync(timeout bool) bool {
 	if ta.SyncFunc == nil || ta.CheckSyncFunc == nil {
 		ta.CanDoReqs = ta.ToDoReqs
-		ta.ToDoReqs = ta.ToDoReqs[0:0]
+		ta.ToDoReqs = nil
 		return true
 	}
 	if !timeout && len(ta.ToDoReqs) == 0 {
@@ -193,7 +198,8 @@ func (ta *Allocator) sync(timeout bool) bool {
 		return false
 	}
 
-	ret := ta.SyncFunc()
+	var ret bool
+	ret, ta.SyncErr = ta.SyncFunc()
 
 	if !timeout {
 		ta.TChan.Reset()
@@ -207,16 +213,28 @@ func (ta *Allocator) finishSyncRequest() {
 			req.Notify(nil)
 		}
 	}
-	ta.SyncReqs = ta.SyncReqs[0:0]
+	ta.SyncReqs = nil
 }
 
 func (ta *Allocator) failRemainRequest() {
+	var err error
+	if ta.SyncErr != nil {
+		err = fmt.Errorf("%s failRemainRequest err:%w", ta.Role, ta.SyncErr)
+	} else {
+		errMsg := fmt.Sprintf("%s failRemainRequest unexpected error", ta.Role)
+		err = errors.New(errMsg)
+	}
+	if len(ta.ToDoReqs) > 0 {
+		log.Debug("Allocator has some reqs to fail",
+			zap.Any("Role", ta.Role),
+			zap.Any("reqLen", len(ta.ToDoReqs)))
+	}
 	for _, req := range ta.ToDoReqs {
 		if req != nil {
-			req.Notify(errors.New("failed: unexpected error"))
+			req.Notify(err)
 		}
 	}
-	ta.ToDoReqs = []Request{}
+	ta.ToDoReqs = nil
 }
 
 func (ta *Allocator) finishRequest() {
@@ -241,7 +259,8 @@ func (ta *Allocator) Close() {
 	ta.CancelFunc()
 	ta.wg.Wait()
 	ta.TChan.Close()
-	ta.revokeRequest(errors.New("closing"))
+	errMsg := fmt.Sprintf("%s is closing", ta.Role)
+	ta.revokeRequest(errors.New(errMsg))
 }
 
 func (ta *Allocator) CleanCache() {

internal/allocator/id.go

@@ -2,14 +2,16 @@ package allocator
 
 import (
 	"context"
-	"log"
+	"fmt"
 	"time"
 
-	"github.com/zilliztech/milvus-distributed/internal/util/retry"
+	"go.uber.org/zap"
 	"google.golang.org/grpc"
 
+	"github.com/zilliztech/milvus-distributed/internal/log"
 	"github.com/zilliztech/milvus-distributed/internal/proto/commonpb"
 	"github.com/zilliztech/milvus-distributed/internal/proto/masterpb"
+	"github.com/zilliztech/milvus-distributed/internal/util/retry"
 	"github.com/zilliztech/milvus-distributed/internal/util/typeutil"
 )
 
@@ -41,6 +43,7 @@ func NewIDAllocator(ctx context.Context, masterAddr string) (*IDAllocator, error
 		Allocator: Allocator{
 			Ctx:        ctx1,
 			CancelFunc: cancel,
+			Role:       "IDAllocator",
 		},
 		countPerRPC:   IDCountPerRPC,
 		masterAddress: masterAddr,
@@ -58,12 +61,11 @@ func (ia *IDAllocator) Start() error {
 	connectMasterFn := func() error {
 		return ia.connectMaster()
 	}
-	err := retry.Retry(10, time.Millisecond*200, connectMasterFn)
+	err := retry.Retry(1000, time.Millisecond*200, connectMasterFn)
 	if err != nil {
 		panic("connect to master failed")
 	}
-	ia.Allocator.Start()
-	return nil
+	return ia.Allocator.Start()
 }
 
 func (ia *IDAllocator) connectMaster() error {
@@ -71,16 +73,31 @@ func (ia *IDAllocator) connectMaster() error {
 	defer cancel()
 	conn, err := grpc.DialContext(ctx, ia.masterAddress, grpc.WithInsecure(), grpc.WithBlock())
 	if err != nil {
-		log.Printf("Connect to master failed, error= %v", err)
+		log.Error("Connect to master failed", zap.Any("Role", ia.Role), zap.Error(err))
 		return err
 	}
-	log.Printf("Connected to master, master_addr=%s", ia.masterAddress)
+	log.Debug("Connected to master", zap.Any("Role", ia.Role), zap.Any("masterAddress", ia.masterAddress))
 	ia.masterConn = conn
 	ia.masterClient = masterpb.NewMasterServiceClient(conn)
 	return nil
 }
 
-func (ia *IDAllocator) syncID() bool {
+func (ia *IDAllocator) gatherReqIDCount() uint32 {
+	need := uint32(0)
+	for _, req := range ia.ToDoReqs {
+		tReq := req.(*IDRequest)
+		need += tReq.count
+	}
+	return need
+}
+
+func (ia *IDAllocator) syncID() (bool, error) {
+
+	need := ia.gatherReqIDCount()
+	if need < ia.countPerRPC {
+		need = ia.countPerRPC
+	}
+
 	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	req := &masterpb.AllocIDRequest{
 		Base: &commonpb.MsgBase{
@@ -89,18 +106,17 @@ func (ia *IDAllocator) syncID() bool {
 			Timestamp: 0,
 			SourceID:  ia.PeerID,
 		},
-		Count: ia.countPerRPC,
+		Count: need,
 	}
 	resp, err := ia.masterClient.AllocID(ctx, req)
 
 	cancel()
 	if err != nil {
-		log.Println("syncID Failed!!!!!")
-		return false
+		return false, fmt.Errorf("syncID Failed:%w", err)
 	}
 	ia.idStart = resp.GetID()
 	ia.idEnd = ia.idStart + int64(resp.GetCount())
-	return true
+	return true, nil
 }
 
 func (ia *IDAllocator) checkSyncFunc(timeout bool) bool {
@@ -122,6 +138,10 @@ func (ia *IDAllocator) pickCanDoFunc() {
 		}
 	}
 	ia.ToDoReqs = ia.ToDoReqs[idx:]
+	log.Debug("IDAllocator pickCanDoFunc",
+		zap.Any("need", need),
+		zap.Any("total", total),
+		zap.Any("remainReqCnt", len(ia.ToDoReqs)))
 }
 
 func (ia *IDAllocator) processFunc(req Request) error {

internal/allocator/timestamp.go

@@ -3,9 +3,11 @@ package allocator
 import (
 	"context"
 	"fmt"
-	"log"
 	"time"
 
+	"go.uber.org/zap"
+
+	"github.com/zilliztech/milvus-distributed/internal/log"
 	"github.com/zilliztech/milvus-distributed/internal/proto/commonpb"
 	"github.com/zilliztech/milvus-distributed/internal/proto/masterpb"
 	"github.com/zilliztech/milvus-distributed/internal/util/retry"
@@ -38,6 +40,7 @@ func NewTimestampAllocator(ctx context.Context, masterAddr string) (*TimestampAl
 		Allocator: Allocator{
 			Ctx:        ctx1,
 			CancelFunc: cancel,
+			Role:       "TimestampAllocator",
 		},
 		masterAddress: masterAddr,
 		countPerRPC:   tsCountPerRPC,
@@ -57,7 +60,7 @@ func (ta *TimestampAllocator) Start() error {
 	connectMasterFn := func() error {
 		return ta.connectMaster()
 	}
-	err := retry.Retry(10, time.Millisecond*200, connectMasterFn)
+	err := retry.Retry(1000, time.Millisecond*200, connectMasterFn)
 	if err != nil {
 		panic("Timestamp local allocator connect to master failed")
 	}
@@ -70,10 +73,10 @@ func (ta *TimestampAllocator) connectMaster() error {
 	defer cancel()
 	conn, err := grpc.DialContext(ctx, ta.masterAddress, grpc.WithInsecure(), grpc.WithBlock())
 	if err != nil {
-		log.Printf("Connect to master failed, error= %v", err)
+		log.Error("TimestampAllocator Connect to master failed", zap.Error(err))
 		return err
 	}
-	log.Printf("Connected to master, master_addr=%s", ta.masterAddress)
+	log.Debug("TimestampAllocator connected to master", zap.Any("masterAddress", ta.masterAddress))
 	ta.masterConn = conn
 	ta.masterClient = masterpb.NewMasterServiceClient(conn)
 	return nil
@@ -98,9 +101,26 @@ func (ta *TimestampAllocator) pickCanDoFunc() {
 		}
 	}
 	ta.ToDoReqs = ta.ToDoReqs[idx:]
+	log.Debug("TimestampAllocator pickCanDoFunc",
+		zap.Any("need", need),
+		zap.Any("total", total),
+		zap.Any("remainReqCnt", len(ta.ToDoReqs)))
 }
 
-func (ta *TimestampAllocator) syncTs() bool {
+func (ta *TimestampAllocator) gatherReqTsCount() uint32 {
+	need := uint32(0)
+	for _, req := range ta.ToDoReqs {
+		tReq := req.(*TSORequest)
+		need += tReq.count
+	}
+	return need
+}
+
+func (ta *TimestampAllocator) syncTs() (bool, error) {
+	need := ta.gatherReqTsCount()
+	if need < ta.countPerRPC {
+		need = ta.countPerRPC
+	}
 	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
 	req := &masterpb.AllocTimestampRequest{
 		Base: &commonpb.MsgBase{
@@ -109,18 +129,18 @@ func (ta *TimestampAllocator) syncTs() bool {
 			Timestamp: 0,
 			SourceID:  ta.PeerID,
 		},
-		Count: ta.countPerRPC,
+		Count: need,
 	}
+
 	resp, err := ta.masterClient.AllocTimestamp(ctx, req)
+	defer cancel()
 
-	cancel()
 	if err != nil {
-		log.Println("syncTimestamp Failed!!!!!")
-		return false
+		return false, fmt.Errorf("syncTimestamp Failed:%w", err)
 	}
 	ta.lastTsBegin = resp.GetTimestamp()
 	ta.lastTsEnd = ta.lastTsBegin + uint64(resp.GetCount())
-	return true
+	return true, nil
 }
 
 func (ta *TimestampAllocator) processFunc(req Request) error {

internal/proxynode/segment.go

@@ -138,6 +138,7 @@ func NewSegIDAssigner(ctx context.Context, dataService types.DataService, getTic
 		Allocator: Allocator{
 			Ctx:        ctx1,
 			CancelFunc: cancel,
+			Role:       "SegmentIDAllocator",
 		},
 		countPerRPC: SegCountPerRPC,
 		dataService: dataService,
@@ -275,9 +276,9 @@ func (sa *SegIDAssigner) reduceSegReqs() {
 	sa.segReqs = newSegReqs
 }
 
-func (sa *SegIDAssigner) syncSegments() bool {
+func (sa *SegIDAssigner) syncSegments() (bool, error) {
 	if len(sa.segReqs) == 0 {
-		return true
+		return true, nil
 	}
 	sa.reduceSegReqs()
 	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
@@ -292,8 +293,7 @@ func (sa *SegIDAssigner) syncSegments() bool {
 	resp, err := sa.dataService.AssignSegmentID(ctx, req)
 
 	if err != nil {
-		log.Debug("proxynode", zap.String("GRPC AssignSegmentID Failed", err.Error()))
-		return false
+		return false, fmt.Errorf("syncSegmentID Failed:%w", err)
 	}
 
 	now := time.Now()
@@ -331,7 +331,7 @@ func (sa *SegIDAssigner) syncSegments() bool {
 		assign.lastInsertTime = now
 		success = true
 	}
-	return success
+	return success, nil
 }
 
 func (sa *SegIDAssigner) processFunc(req allocator.Request) error {