add the deadline for the allocating requests

daemon/daemon.go

@@ -190,7 +190,7 @@
 		if pod.PodENI {
 			resourceRequests = append(resourceRequests, &eni.RemoteIPRequest{})
 		} else {
-			req := &eni.LocalIPRequest{}
+			req := eni.NewLocalIPRequest()
 			if pod.ERdma {
 				req.LocalIPType = eni.LocalIPTypeERDMA
 			}
@@ -208,7 +208,7 @@
 		if pod.PodENI || n.ipamType == types.IPAMTypeCRD {
 			resourceRequests = append(resourceRequests, &eni.RemoteIPRequest{})
 		} else {
-			req := &eni.LocalIPRequest{}
+			req := eni.NewLocalIPRequest()
 
 			if len(oldRes.GetResourceItemByType(daemon.ResourceTypeENI)) == 1 {
 				old := oldRes.GetResourceItemByType(daemon.ResourceTypeENI)[0]

pkg/eni/local.go

@@ -11,6 +11,8 @@
 	"sync"
 	"time"
 
+	"github.com/go-logr/logr"
+	"github.com/samber/lo"
 	"golang.org/x/time/rate"
 	corev1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/util/cache"
@@ -51,13 +53,24 @@
 
 var _ ResourceRequest = &LocalIPRequest{}
 
+func NewLocalIPRequest() *LocalIPRequest {
+	ctx, cancel := context.WithCancel(context.Background())
+	return &LocalIPRequest{
+		workerCtx: ctx,
+		cancel:    cancel,
+	}
+}
+
 type LocalIPRequest struct {
 	NetworkInterfaceID string
 	LocalIPType        string
 	IPv4               netip.Addr
 	IPv6               netip.Addr
 
 	NoCache bool // do not use cached ip
+
+	workerCtx context.Context
+	cancel    context.CancelFunc
 }
 
 func (l *LocalIPRequest) ResourceType() ResourceType {
@@ -122,7 +135,9 @@
 	batchSize int
 
 	cap                        int
-	allocatingV4, allocatingV6 int
+	allocatingV4, allocatingV6 AllocatingRequests
+	// danging, used for release
+	dangingV4, dangingV6 AllocatingRequests
 
 	eni                    *daemon.ENI
 	ipAllocInhibitExpireAt time.Time
@@ -375,30 +390,30 @@
 		return nil, nil
 	}
 
-	lo, ok := request.(*LocalIPRequest)
+	localIPRequest, ok := request.(*LocalIPRequest)
 	if !ok {
 		return nil, []Trace{{Condition: ResourceTypeMismatch}}
 	}
 
-	if lo.NetworkInterfaceID != "" && l.eni != nil && l.eni.ID != lo.NetworkInterfaceID {
+	if localIPRequest.NetworkInterfaceID != "" && l.eni != nil && l.eni.ID != localIPRequest.NetworkInterfaceID {
 		return nil, []Trace{{Condition: NetworkInterfaceMismatch}}
 	}
 
-	log := logf.FromContext(ctx)
-	log.Info(fmt.Sprintf("local request %v", lo))
+	log := logr.FromContextOrDiscard(ctx)
+	log.Info(fmt.Sprintf("local request %v", localIPRequest))
 
 	expectV4 := 0
 	expectV6 := 0
 
 	if l.enableIPv4 {
-		if lo.NoCache {
-			if len(l.ipv4)+l.allocatingV4 >= l.cap {
+		if localIPRequest.NoCache {
+			if len(l.ipv4)+l.allocatingV4.Len() >= l.cap {
 				return nil, []Trace{{Condition: Full}}
 			}
 			expectV4 = 1
 		} else {
 			ipv4 := l.ipv4.PeekAvailable(cni.PodID)
-			if ipv4 == nil && len(l.ipv4)+l.allocatingV4 >= l.cap {
+			if ipv4 == nil && len(l.ipv4)+l.allocatingV4.Len() >= l.cap {
 				return nil, []Trace{{Condition: Full}}
 			} else if ipv4 == nil {
 				expectV4 = 1
@@ -407,14 +422,14 @@
 	}
 
 	if l.enableIPv6 {
-		if lo.NoCache {
-			if len(l.ipv6)+l.allocatingV6 >= l.cap {
+		if localIPRequest.NoCache {
+			if len(l.ipv6)+l.allocatingV6.Len() >= l.cap {
 				return nil, []Trace{{Condition: Full}}
 			}
 			expectV6 = 1
 		} else {
 			ipv6 := l.ipv6.PeekAvailable(cni.PodID)
-			if ipv6 == nil && len(l.ipv6)+l.allocatingV6 >= l.cap {
+			if ipv6 == nil && len(l.ipv6)+l.allocatingV6.Len() >= l.cap {
 				return nil, []Trace{{Condition: Full}}
 			} else if ipv6 == nil {
 				expectV6 = 1
@@ -427,21 +442,18 @@
 		return nil, []Trace{{Condition: InsufficientVSwitchIP, Reason: fmt.Sprintf("alloc inhibit, expire at %s", l.ipAllocInhibitExpireAt.String())}}
 	}
 
-	l.allocatingV4 += expectV4
-	l.allocatingV6 += expectV6
+	for i := 0; i < expectV4; i++ {
+		l.allocatingV4 = append(l.allocatingV4, localIPRequest)
+	}
+	for i := 0; i < expectV6; i++ {
+		l.allocatingV6 = append(l.allocatingV6, localIPRequest)
+	}
 
 	l.cond.Broadcast()
 
 	respCh := make(chan *AllocResp)
 
-	go l.allocWorker(ctx, cni, lo, respCh, func() {
-		// current roll back ip at same time
-		l.allocatingV4 -= expectV4
-		l.allocatingV4 = max(l.allocatingV4, 0)
-		l.allocatingV6 -= expectV6
-		l.allocatingV6 = max(l.allocatingV6, 0)
-		log.Info("rollback ipv4", "ipv4", expectV4)
-	})
+	go l.allocWorker(ctx, cni, localIPRequest, respCh)
 
 	return respCh, nil
 }
@@ -461,7 +473,7 @@
 		return false, nil
 	}
 
-	log := logf.FromContext(ctx)
+	log := logr.FromContextOrDiscard(ctx)
 
 	if res.IP.IPv4.IsValid() {
 		l.ipv4.Release(cni.PodID, res.IP.IPv4)
@@ -509,13 +521,24 @@
 }
 
 // allocWorker started with each Allocate call
-func (l *Local) allocWorker(ctx context.Context, cni *daemon.CNI, request *LocalIPRequest, respCh chan *AllocResp, onErrLocked func()) {
+func (l *Local) allocWorker(ctx context.Context, cni *daemon.CNI, request *LocalIPRequest, respCh chan *AllocResp) {
 	done := make(chan struct{})
 	defer close(done)
 
 	l.cond.L.Lock()
 	defer l.cond.L.Unlock()
 
+	defer func() {
+		if request == nil {
+			return
+		}
+
+		l.switchIPv4(request)
+		l.switchIPv6(request)
+
+		request.cancel()
+	}()
+
 	go func() {
 		select {
 		case <-ctx.Done():
@@ -526,13 +549,11 @@
 		}
 	}()
 
-	log := logf.FromContext(ctx)
+	log := logr.FromContextOrDiscard(ctx)
 	for {
 		select {
 		case <-ctx.Done():
 			// parent cancel the context, so close the ch
-			onErrLocked()
-
 			close(respCh)
 			return
 		default:
@@ -592,7 +613,7 @@
 func (l *Local) factoryAllocWorker(ctx context.Context) {
 	l.cond.L.Lock()
 
-	log := logf.FromContext(ctx)
+	log := logr.FromContextOrDiscard(ctx)
 	for {
 		if log.V(4).Enabled() {
 			log.V(4).Info("call allocWorker")
@@ -605,7 +626,7 @@
 		default:
 		}
 
-		if l.allocatingV4 <= 0 && l.allocatingV6 <= 0 {
+		if l.allocatingV4.Len() <= 0 && l.allocatingV6.Len() <= 0 {
 			l.cond.Wait()
 			continue
 		}
@@ -629,8 +650,8 @@
 
 		if l.eni == nil {
 			// create eni
-			v4Count := min(l.batchSize, max(l.allocatingV4, 1))
-			v6Count := min(l.batchSize, l.allocatingV6)
+			v4Count := min(l.batchSize, max(l.allocatingV4.Len(), 1))
+			v6Count := min(l.batchSize, l.allocatingV6.Len())
 
 			l.status = statusCreating
 			l.cond.L.Unlock()
@@ -669,11 +690,8 @@
 
 			l.eni = eni
 
-			l.allocatingV4 -= v4Count
-			l.allocatingV6 -= v6Count
-
-			l.allocatingV4 = max(l.allocatingV4, 0)
-			l.allocatingV6 = max(l.allocatingV6, 0)
+			l.popNIPv4Jobs(v4Count)
+			l.popNIPv6Jobs(v6Count)
 
 			primary, err := netip.ParseAddr(eni.PrimaryIP.IPv4.String())
 			if err == nil {
@@ -693,8 +711,8 @@
 			l.status = statusInUse
 		} else {
 			eniID := l.eni.ID
-			v4Count := min(l.batchSize, l.allocatingV4)
-			v6Count := min(l.batchSize, l.allocatingV6)
+			v4Count := min(l.batchSize, l.allocatingV4.Len())
+			v6Count := min(l.batchSize, l.allocatingV6.Len())
 
 			if v4Count > 0 {
 				l.cond.L.Unlock()
@@ -718,8 +736,7 @@
 					continue
 				}
 
-				l.allocatingV4 -= len(ipv4Set)
-				l.allocatingV4 = max(l.allocatingV4, 0)
+				l.popNIPv4Jobs(len(ipv4Set))
 
 				l.ipv4.PutValid(ipv4Set...)
 
@@ -751,8 +768,7 @@
 					continue
 				}
 
-				l.allocatingV6 -= len(ipv6Set)
-				l.allocatingV6 = max(l.allocatingV6, 0)
+				l.popNIPv6Jobs(len(ipv6Set))
 
 				l.ipv6.PutValid(ipv6Set...)
 
@@ -841,7 +857,7 @@
 func (l *Local) factoryDisposeWorker(ctx context.Context) {
 	l.cond.L.Lock()
 
-	log := logf.FromContext(ctx)
+	log := logr.FromContextOrDiscard(ctx)
 	for {
 		select {
 		case <-ctx.Done():
@@ -1062,3 +1078,68 @@
 	}
 	return parts[0], parts[1], nil
 }
+
+func (l *Local) switchIPv4(req *LocalIPRequest) {
+	found := false
+	l.allocatingV4 = lo.Filter(l.allocatingV4, func(item *LocalIPRequest, index int) bool {
+		if item != req {
+			// true to keep
+			return true
+		}
+		found = true
+		return false
+	})
+	if !found {
+		return
+	}
+
+	if l.dangingV4.Len() == 0 {
+		// this may not happen
+		// call the Len() to make sure canceled job will be removed
+		return
+	}
+	l.allocatingV4 = append(l.allocatingV4, l.dangingV4[0])
+	l.dangingV4 = l.dangingV4[1:]
+}
+
+func (l *Local) switchIPv6(req *LocalIPRequest) {
+	found := false
+	l.allocatingV6 = lo.Filter(l.allocatingV6, func(item *LocalIPRequest, index int) bool {
+		if item != req {
+			// true to keep
+			return true
+		}
+		found = true
+		return false
+	})
+	if !found {
+		return
+	}
+
+	if l.dangingV6.Len() == 0 {
+		// this may not happen
+		return
+	}
+	l.allocatingV6 = append(l.allocatingV6, l.dangingV6[0])
+	l.dangingV6 = l.dangingV6[1:]
+}
+
+func (l *Local) popNIPv4Jobs(count int) {
+	firstPart, secondPart := Split(l.allocatingV4, count)
+	l.dangingV4 = append(l.dangingV4, firstPart...)
+	l.allocatingV4 = secondPart
+}
+
+func (l *Local) popNIPv6Jobs(count int) {
+	firstPart, secondPart := Split(l.allocatingV6, count)
+	l.dangingV6 = append(l.dangingV6, firstPart...)
+	l.allocatingV6 = secondPart
+}
+
+func Split[T any](arr []T, index int) ([]T, []T) {
+	if index < 0 || index > len(arr) {
+		return arr, nil
+	}
+
+	return arr[:index], arr[index:]
+}