refactor: standardize variable naming for memory and lock management

Author: haitwang-cloud

pkg/nvidia-plugin/pkg/plugin/register.go

@@ -137,11 +137,11 @@ func (plugin *NvidiaDevicePlugin) getAPIDevices() *[]*util.DeviceInfo {
 			panic(0)
 		}
 
-		registeredmem := int32(memoryTotal / 1024 / 1024)
+		registeredMem := int32(memoryTotal / 1024 / 1024)
 		if plugin.schedulerConfig.DeviceMemoryScaling != 1 {
-			registeredmem = int32(float64(registeredmem) * plugin.schedulerConfig.DeviceMemoryScaling)
+			registeredMem = int32(float64(registeredMem) * plugin.schedulerConfig.DeviceMemoryScaling)
 		}
-		klog.Infoln("MemoryScaling=", plugin.schedulerConfig.DeviceMemoryScaling, "registeredmem=", registeredmem)
+		klog.Infoln("MemoryScaling=", plugin.schedulerConfig.DeviceMemoryScaling, "registeredMem=", registeredMem)
 		health := true
 		for _, val := range devs {
 			if strings.Compare(val.ID, UUID) == 0 {
@@ -163,14 +163,14 @@ func (plugin *NvidiaDevicePlugin) getAPIDevices() *[]*util.DeviceInfo {
 			ID:      UUID,
 			Index:   uint(idx),
 			Count:   int32(plugin.schedulerConfig.DeviceSplitCount),
-			Devmem:  registeredmem,
+			Devmem:  registeredMem,
 			Devcore: int32(plugin.schedulerConfig.DeviceCoreScaling * 100),
 			Type:    fmt.Sprintf("%v-%v", "NVIDIA", Model),
 			Numa:    numa,
 			Mode:    plugin.operatingMode,
 			Health:  health,
 		})
-		klog.Infof("nvml registered device id=%v, memory=%v, type=%v, numa=%v", idx, registeredmem, Model, numa)
+		klog.Infof("nvml registered device id=%v, memory=%v, type=%v, numa=%v", idx, registeredMem, Model, numa)
 	}
 	return &res
 }
@@ -184,9 +184,9 @@ func (plugin *NvidiaDevicePlugin) RegistrInAnnotation() error {
 		klog.Errorln("get node error", err.Error())
 		return err
 	}
-	encodeddevices := util.EncodeNodeDevices(*devices)
+	encodedDevices := util.EncodeNodeDevices(*devices)
 	annos[nvidia.HandshakeAnnos] = "Reported " + time.Now().String()
-	annos[nvidia.RegisterAnnos] = encodeddevices
+	annos[nvidia.RegisterAnnos] = encodedDevices
 	klog.Infof("patch node with the following annos %v", fmt.Sprintf("%v", annos))
 	err = util.PatchNodeAnnotations(node, annos)
 

pkg/scheduler/scheduler.go

@@ -420,7 +420,7 @@ func (s *Scheduler) Bind(args extenderv1.ExtenderBindingArgs) (*extenderv1.Exten
 		return res, nil
 	}
 
-	tmppatch := map[string]string{
+	deviceBindAnnotations := map[string]string{
 		util.DeviceBindPhase:     "allocating",
 		util.BindTimeAnnotations: strconv.FormatInt(time.Now().Unix(), 10),
 	}
@@ -433,7 +433,7 @@ func (s *Scheduler) Bind(args extenderv1.ExtenderBindingArgs) (*extenderv1.Exten
 		}
 	}
 
-	err = util.PatchPodAnnotations(current, tmppatch)
+	err = util.PatchPodAnnotations(current, deviceBindAnnotations)
 	if err != nil {
 		klog.ErrorS(err, "Failed to patch pod annotations", "pod", klog.KObj(current))
 		return &extenderv1.ExtenderBindingResult{Error: err.Error()}, err

pkg/util/nodelock/nodelock.go

@@ -42,7 +42,7 @@ const (
 
 var lock sync.Mutex
 
-func SetNodeLock(nodeName string, lockname string, pods *corev1.Pod) error {
+func SetNodeLock(nodeName string, lockName string, pods *corev1.Pod) error {
 	lock.Lock()
 	defer lock.Unlock()
 	ctx := context.Background()
@@ -80,7 +80,7 @@ func SetNodeLock(nodeName string, lockname string, pods *corev1.Pod) error {
 	return nil
 }
 
-func ReleaseNodeLock(nodeName string, lockname string, pod *corev1.Pod, timeout bool) error {
+func ReleaseNodeLock(nodeName string, lockName string, pod *corev1.Pod, timeout bool) error {
 	lock.Lock()
 	defer lock.Unlock()
 	ctx := context.Background()
@@ -118,27 +118,37 @@ func ReleaseNodeLock(nodeName string, lockname string, pod *corev1.Pod, timeout
 	return nil
 }
 
-func LockNode(nodeName string, lockname string, pods *corev1.Pod) error {
+func LockNode(nodeName string, lockName string, pods *corev1.Pod) error {
 	ctx := context.Background()
 	node, err := client.GetClient().CoreV1().Nodes().Get(ctx, nodeName, metav1.GetOptions{})
 	if err != nil {
 		return err
 	}
 	if _, ok := node.ObjectMeta.Annotations[NodeLockKey]; !ok {
-		return SetNodeLock(nodeName, lockname, pods)
+		return SetNodeLock(nodeName, lockName, pods)
 	}
 	lockTime, _, _, err := ParseNodeLock(node.ObjectMeta.Annotations[NodeLockKey])
 	if err != nil {
 		return err
 	}
+	klog.InfoS("Attempting to lock node", "node", nodeName, "pod", pods.Name)
+	if _, ok := node.ObjectMeta.Annotations[NodeLockKey]; !ok {
+		klog.InfoS("No existing lock found", "node", nodeName)
+	} else {
+		klog.InfoS("Existing lock details",
+			"node", nodeName,
+			"lockTime", lockTime,
+			"currentTime", time.Now(),
+			"timeSinceLock", time.Since(lockTime))
+	}
 	if time.Since(lockTime) > time.Minute*5 {
 		klog.InfoS("Node lock expired", "node", nodeName, "lockTime", lockTime)
-		err = ReleaseNodeLock(nodeName, lockname, pods, true)
+		err = ReleaseNodeLock(nodeName, lockName, pods, true)
 		if err != nil {
 			klog.ErrorS(err, "Failed to release node lock", "node", nodeName)
 			return err
 		}
-		return SetNodeLock(nodeName, lockname, pods)
+		return SetNodeLock(nodeName, lockName, pods)
 	}
 	return fmt.Errorf("node %s has been locked within 5 minutes", nodeName)
 }