kubernetes - 基于集群中节点数量的水平Pod Autoscaler副本

Question

我正在寻找一种解决方案，该解决方案将在节点加入群集时自动扩展Pod，并在删除节点时自动扩展。
我们正在节点上运行WebApp，当计划断开该节点的连接时，这需要正常的pod逐出/终止。
我正在检查使用DaemonSet的选项，但是由于我们使用Kops进行群集滚动更新，因此它忽略了DaemonSets驱逐(不支持标志“--ignore-daemionset”)。
结果，WebApp与该节点“死亡”，这对于我们的应用程序是 Not Acceptable 。
Horizo​​ntalPodAutoscaler能够覆盖部署yaml中设置的副本数量的功能可以解决此问题。
我想找到一种基于集群中节点数量动态更改Horizo​​ntalPodAutoscaler yaml中的min / maxReplicas的方法。

spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: MyWebApp
  minReplicas: "Num of nodes in the cluster"
  maxReplicas: "Num of nodes in the cluster"

有什么想法如何获取节点数并相应地更新集群中的Horizo​​ntalPodAutoscaler yaml？或任何其他解决方案？

Answer 1

您是否尝试过在daemonset yaml中使用nodeSelector规范。
因此，如果您在yaml中设置了nodeselector，则；如果从节点中删除nodeselector标签值，则在耗尽之前，守护进程应适当缩小。也是如此，当您将新节点添加到集群标签中时，带有自定义值的节点和deamonset将会放大。

这对我有效，因此您可以尝试并与Kops确认

首先:使用将始终在集群上拥有的自定义标签来标记所有节点

例:

kubectl label nodes k8s-master-1 mylabel=allow_demon_set  
kubectl label nodes k8s-node-1 mylabel=allow_demon_set
kubectl label nodes k8s-node-2 mylabel=allow_demon_set
kubectl label nodes k8s-node-3 mylabel=allow_demon_set

然后在守护进程中设置yaml并添加节点选择器。

Example.yaml的用法如下: 注意 添加了节点elctor字段

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: fluentd-elasticsearch
  labels:
    k8s-app: fluentd-logging
spec:
  selector:
    matchLabels:
      name: fluentd-elasticsearch
  template:
    metadata:
      labels:
        name: fluentd-elasticsearch
    spec:
      nodeSelector:
        mylabel: allow_demon_set
      tolerations:
      - key: node-role.kubernetes.io/master
        effect: NoSchedule
      containers:
      - name: fluentd-elasticsearch
        image: quay.io/fluentd_elasticsearch/fluentd:v2.5.2
        resources:
          limits:
            memory: 200Mi
          requests:
            cpu: 100m
            memory: 200Mi
        volumeMounts:
        - name: varlog
          mountPath: /var/log
        - name: varlibdockercontainers
          mountPath: /var/lib/docker/containers
          readOnly: true
      terminationGracePeriodSeconds: 30
      volumes:
      - name: varlog
        hostPath:
          path: /var/log
      - name: varlibdockercontainers
        hostPath:
          path: /var/lib/docker/containers

所以节点被标记如下

$ kubectl get nodes --show-labels
NAME           STATUS   ROLES    AGE   VERSION   LABELS
k8s-master-1   Ready    master   9d    v1.17.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-master-1,kubernetes.io/os=linux,mylable=allow_demon_set,node-role.kubernetes.io/master=
k8s-node-1     Ready    <none>   9d    v1.17.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node-1,kubernetes.io/os=linux,mylable=allow_demon_set
k8s-node-2     Ready    <none>   9d    v1.17.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node-2,kubernetes.io/os=linux,mylable=allow_demon_set
k8s-node-3     Ready    <none>   9d    v1.17.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node-3,kubernetes.io/os=linux,mylable=allow_demon_set

正确的Yaml后，启动使用它的守护程序集

$ kubectl create -f Example.yaml

$ kubectl get all -o wide
NAME                              READY   STATUS    RESTARTS   AGE   IP            NODE           NOMINATED NODE   READINESS GATES
pod/fluentd-elasticsearch-jrgl6   1/1     Running   0          20s   10.244.3.19   k8s-node-3     <none>           <none>
pod/fluentd-elasticsearch-rgcm2   1/1     Running   0          20s   10.244.0.6    k8s-master-1   <none>           <none>
pod/fluentd-elasticsearch-wccr9   1/1     Running   0          20s   10.244.1.14   k8s-node-1     <none>           <none>
pod/fluentd-elasticsearch-wxq5v   1/1     Running   0          20s   10.244.2.33   k8s-node-2     <none>           <none>

NAME                 TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE   SELECTOR
service/kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   9d    <none>

NAME                                   DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR             AGE   CONTAINERS              IMAGES                                         SELECTOR
daemonset.apps/fluentd-elasticsearch   4         4         4       4            4           mylable=allow_demon_set   20s   fluentd-elasticsearch   quay.io/fluentd_elasticsearch/fluentd:v2.5.2   name=fluentd-elasticsearch

然后在耗尽节点之前添加 ，我们可以从节点中删除自定义标签，并且pod应当适当缩小比例，然后耗尽节点。

$ kubectl label nodes k8s-node-3 mylabel-

检查守护程序集，它应按比例缩小

ubuntu@k8s-kube-client:~$ kubectl get all -o wide
NAME                              READY   STATUS        RESTARTS   AGE     IP            NODE           NOMINATED NODE   READINESS GATES
pod/fluentd-elasticsearch-jrgl6   0/1     Terminating   0          2m36s   10.244.3.19   k8s-node-3     <none>           <none>
pod/fluentd-elasticsearch-rgcm2   1/1     Running       0          2m36s   10.244.0.6    k8s-master-1   <none>           <none>
pod/fluentd-elasticsearch-wccr9   1/1     Running       0          2m36s   10.244.1.14   k8s-node-1     <none>           <none>
pod/fluentd-elasticsearch-wxq5v   1/1     Running       0          2m36s   10.244.2.33   k8s-node-2     <none>           <none>

NAME                 TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE   SELECTOR
service/kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   9d    <none>

NAME                                   DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR             AGE     CONTAINERS              IMAGES                                         SELECTOR
daemonset.apps/fluentd-elasticsearch   3         3         3       3            3           mylable=allow_demon_set   2m36s   fluentd-elasticsearch   quay.io/fluentd_elasticsearch/fluentd:v2.5.2   name=fluentd-elasticsearch

现在，再次将 标签添加到具有相同自定义标签的新节点时，将其添加到集群中，并且deamonset将按比例放大

$ kubectl label nodes k8s-node-3 mylable=allow_demon_set

ubuntu@k8s-kube-client:~$ kubectl get all -o wide
NAME                              READY   STATUS    RESTARTS   AGE     IP            NODE           NOMINATED NODE   READINESS GATES
pod/fluentd-elasticsearch-22rsj   1/1     Running   0          2s      10.244.3.20   k8s-node-3     <none>           <none>
pod/fluentd-elasticsearch-rgcm2   1/1     Running   0          5m28s   10.244.0.6    k8s-master-1   <none>           <none>
pod/fluentd-elasticsearch-wccr9   1/1     Running   0          5m28s   10.244.1.14   k8s-node-1     <none>           <none>
pod/fluentd-elasticsearch-wxq5v   1/1     Running   0          5m28s   10.244.2.33   k8s-node-2     <none>           <none>

NAME                 TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE   SELECTOR
service/kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   9d    <none>

NAME                                   DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR             AGE     CONTAINERS              IMAGES                                         SELECTOR
daemonset.apps/fluentd-elasticsearch   4         4         4       4            4           mylable=allow_demon_set   5m28s   fluentd-elasticsearch   quay.io/fluentd_elasticsearch/fluentd:v2.5.2   name=fluentd-elasticsearch

请确认这是您想做的并与kops一起使用