基于集群中节点数量的水平Pod Autoscaler副本

我正在寻找一种解决方案，当节点加入集群时将自动扩展pod，并在删除节点时将其自动扩展。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不可接受的。 HorizontalpodAutoscaler能够覆盖在部署Yaml中设置的副本数量的功能可以解决此问题。我想找到一种方法，可以根据集群中的节点数量动态地更改HorizontalpodAutoscaler yaml中的min / maxReplicas。

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HorizontalpodAutoscaler yaml？或任何其他解决方案？

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，并且在从流失前删除节点选择器标签值，则守护进程应适当缩小。在将新节点添加到带有自定义值的集群标签中时也是如此和恶魔会扩大规模。

这对我有用，因此您可以尝试一下并与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

示例：

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

然后将其设置为守护程序yaml添加节点选择器。

Example.yaml的用法如下：注意 添加了nodeselctor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

因此，节点标记如下

$ 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,kubernetes.io/hostname=k8s-node-1,mylable=allow_demon_set
k8s-node-2     Ready    <none>   9d    v1.17.0   beta.kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node-2,mylable=allow_demon_set
k8s-node-3     Ready    <none>   9d    v1.17.0   beta.kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node-3,mylable=allow_demon_set

一旦您使用了正确的Yaml，就启动使用它的守护进程集

$ 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，我们只需从节点上删除自定义标签，并且pod应当适当缩小，然后耗尽节点。

$ kubectl label nodes k8s-node-3 mylabel-

检查后台驻留程序，它应该缩小

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，在将标签添加到集群时，再次将标签添加到具有相同自定义标签的新节点，并且守护进程将扩大

$ 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

请确认是否要执行此操作并与kops配合使用

基于集群中节点数量的水平Pod Autoscaler副本

whmyth 回答：基于集群中节点数量的水平Pod Autoscaler副本

大家都在问