k8s 通过node节点获取资源信息

• 获取node节点GPU/CPU/Mem 的capacity和request 数

func GetNodeResource(allPods []v1.Pod, node *v1.Node) (resource map[string]model.ResourceStatus, err error) {
    var (
        allocatedGPU int64
        allocatedCPU int64
        allocatedMem int64
        capacityGPU  int64
    )
    resource = map[string]model.ResourceStatus{}
    val, ok := node.Status.Capacity[common.NvidiaGPUResource]
    if !ok {
        capacityGPU = 0
    } else {
        capacityGPU = val.Value()
    }
    for _, pod := range getPodsFromNode(allPods, node) {
        allocatedGPU += gpuInPod(&pod)
        allocatedCPU += cpuInPod(&pod)
        allocatedMem += memInPod(&pod)
    }
    resource[common.ResourceGPU] = model.ResourceStatus{
        Request:  allocatedGPU,
        Capacity: capacityGPU,
    }
    resource[common.ResourceCPU] = model.ResourceStatus{
        Request:  allocatedCPU,
        Capacity: node.Status.Capacity.Cpu().Value(),
    }
    resource[common.ResourceMemory] = model.ResourceStatus{
        Request:  allocatedMem,
        Capacity: node.Status.Capacity.Memory().Value(),
    }
    return resource, nil
}
// 获取节点上的pod
func getPodsFromNode(allPods []v1.Pod, node *v1.Node) (pods []v1.Pod) {
    for _, pod := range allPods {
        if pod.Spec.NodeName == node.Name {
            pods = append(pods, pod)
        }
    }
    return pods
}
// 获取所有的running的pod 
func AllActivePods(clientSet *kubernetes.Clientset) ([]v1.Pod, error) {
    allPods, err := clientSet.CoreV1().Pods("").List(context.TODO(), metav1.ListOptions{
        FieldSelector: "status.phase=Running",
    })
    if err != nil {
        log.Infof("get pods err, %s", err)
        return nil, err
    }
    log.Infof("Running pod count %d", len(allPods.Items))
    return allPods.Items, nil
}
// 获取pod上的gpu request
func gpuInPod(pod *v1.Pod) (gpuCount int64) {
    containers := pod.Spec.Containers
    for _, container := range containers {
        val, ok := container.Resources.Limits[common.NvidiaGPUResource]
        if !ok {
            continue
        }
        gpuCount += val.Value()
    }
    return gpuCount
}
// 获取pod上的cpu request
func cpuInPod(pod *v1.Pod) (cpuCount int64) {
    containers := pod.Spec.Containers
    for _, container := range containers {
        val, ok := container.Resources.Requests[common.ResourceCPU]
        if !ok {
            continue
        }
        cpuCount += val.Value()
    }
    return cpuCount
}
// 获取pod上的mem request
func memInPod(pod *v1.Pod) (memCount int64) {
    containers := pod.Spec.Containers
    for _, container := range containers {
        val, ok := container.Resources.Requests[common.ResourceMemory]
        if !ok {
            continue
        }
        memCount += val.Value()
    }
    return memCount
}

func GetNodeRole(node *v1.Node) string {
    if _, ok := node.Labels[common.MasterLabelRole]; ok {
        return "master"
    }
    return "node"
}