86、pod部署策略

一、集群的调度

集群的调度：

怎么把pod部署到节点的方法。

1.1、调度的过程：

scheduler是集群的调度器，主要任务就是把pod部署到节点上。

1.2、自动调度：

1、公平，保证每个可用的节点都可以部署pod

2、资源的高效利用，集群当中的资源可用被最大化的使用

3、效率，调度的性能要好，能够对大批量的pod完成调度工作。

4、灵活（自定义），用户需要根据自己的需求进行控制，可以满足。

1.3、调度约束机制：

list-watch机制来进行每个组件的协作，保持数据同步，组件之间的解耦。

list-watch

watch：k8s当中的监听

get：获取资源

apiserver和组件之间的watch监听机制：

在这里插入图片描述

二、调度过程：

2.1、默认算法有两种：

1、预算策略

1、预算策略：先对节点的条件进行过滤

pod的资源适应性：节点是否有资源能够满足pod请求的资源

pod的主机适应性：如果指定了节点，检查集群中是否有满足要求的节点可供部署

pod的主机端口适应性：检查节点上使用的端口是否与pod请求的端口冲突

pod与主机磁盘的适应性：每个pod的挂载卷不能冲突。

如果预算条件不满足，pod会进行pending状态

2、优选策略

2、优选策略：根据过滤条件选择一条最优的节点

最低请求优先级：通过计算节点上cpu和内存的使用率，确定节点的权重。使用率越低权重越大，越会被选中作为部署节点。

倾向于选择资源占用较少的节点。

平衡资源分配：cpu和内存的使用率，确定节点的权重。按cpu和内存之间比率，使用率的比率。cpu和内存比越接近1。

A 50% 50% 优先级高

B 10% 20%

镜像本地性优先级：如果节点在本地已经有了需要的镜像，分配的概率更大。

随调度部署nginx1

[root@master01 k8s-yaml]# vim test11.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
        
[root@master01 k8s-yaml]# kubectl apply -f test11.yaml 
deployment.apps/nginx1 created
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-d7tdz   1/1     Running   0          12s   10.244.2.137   node02   <none>           <none>
nginx1-654cb56c4-mp8zz   1/1     Running   0          14s   10.244.2.136   node02   <none>           <none>
nginx1-654cb56c4-qnlxw   1/1     Running   0          13s   10.244.1.98    node01   <none>           <none>

2.2、用户定制节点部署：（重要）

1、强制性节点调度：

nodeName强制性的选择一个节点，不再需要调度器和算法了。之间部署即可。

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,test1=a,test2=b,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b

强制性布置在node01

[root@master01 k8s-yaml]# vim test11.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      nodeName: node01
      
 [root@master01 k8s-yaml]# kubectl apply -f test11.yaml --force
     
      
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-95bf57c7f-nrbjb   1/1     Running   0          48s   10.244.1.101   node01   <none>           <none>
nginx1-95bf57c7f-phxx5   1/1     Running   0          50s   10.244.1.99    node01   <none>           <none>
nginx1-95bf57c7f-zl78g   1/1     Running   0          49s   10.244.1.100   node01   <none>           <none>

2.3、查看节点的标签

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
标签的格式是键值对
一个节点可以有多个标签，每个以逗号隔开

[root@master01 k8s-yaml]# kubectl label nodes node01 test=a      ##增加标签
node/node01 labeled

[root@master01 k8s-yaml]# kubectl label nodes node01 test=b --overwrite   ##改标签
node/node01 labeled

[root@master01 k8s-yaml]# kubectl label nodes node01 test-         ##删除标签
node/node01 labeled

自定义标签

[root@master01 k8s-yaml]# kubectl label nodes node01 test1=a
node/node01 labeled
[root@master01 k8s-yaml]# kubectl label nodes node02 test2=b
node/node02 labeled

2.4、根据节点的标签来进行部署。匹配机制，只要标签匹配都可以部署

根据节点的标签来进行部署。匹配机制，只要标签匹配都可以部署。

问题:标签选择节点是否需要调度器和算法

需要调度器和算法来进行分配的。

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      nodeSelector:
        test2: b
         


[root@master01 k8s-yaml]# kubectl label nodes node01 test3=b
node/node01 labeled
[root@master01 k8s-yaml]# kubectl apply -f test12.yaml 
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-76bd776789-74j2f   1/1     Running   0          21s   10.244.2.138   node02   <none>           <none>
nginx1-76bd776789-qkjtc   1/1     Running   0          19s   10.244.2.139   node02   <none>           <none>
nginx1-76bd776789-z7885   1/1     Running   0          18s   10.244.2.140   node02   <none>           <none>




[root@master01 k8s-yaml]# kubectl label nodes node01 test2=b

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      nodeSelector:
        test2: b
[root@master01 k8s-yaml]# kubectl apply -f test12.yaml 
[root@master01 k8s-yaml]# kubectl get pod -o wide
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-76bd776789-5dssm   1/1     Running   0          9s    10.244.2.141   node02   <none>           <none>
nginx1-76bd776789-mpk42   1/1     Running   0          9s    10.244.1.102   node01   <none>           <none>
nginx1-76bd776789-zpbkp   1/1     Running   0          9s    10.244.2.142   node02   <none>           <none>


##同时都有自定义标签，随机部署在两个节点上

三、亲和性

节点亲和性：node Affinity

pod亲和性：pod Affinity

软策略和硬策略：

软策略:

preferredDuringSchedulingIgnoredDuringExecution

软策略在选择节点尽量满足部署的条件，非条件也可以部署。

硬策略：

requiredDuringSchedulingIgnoredDuringExecution

硬策略：必须满足指定的节点条件，否则pending

1、条件不满足，肯定pending

2、条件满足，调度器即可生效

3、需要调度器的分配，不同节点可以有相同的标签。需要调度器分配。

根据节点标签和pod的标签来进行选择：

键值的运算关系：

1、In 在  匹配 =

2、Notin 不在 不等于  ，逻辑非

3、Gt 大于

4、Lt 小于

5、Exists 存在

6、DoesNotExist 不存在

3.1、亲和性：

topologkey 定义节点的拓扑域，用来反映pod和节点之间的关系。

硬策略

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: test1
                operator: In
                values:
                - a
#节点亲和性的硬策略，表示必须选择带有标签的值是test1=a
[root@master01 k8s-yaml]# kubectl apply -f test12.yaml 

[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-8f7b9d885-4ds5z   1/1     Running   0          21s   10.244.1.103   node01   <none>           <none>
nginx1-8f7b9d885-c9zbc   1/1     Running   0          19s   10.244.1.105   node01   <none>           <none>
nginx1-8f7b9d885-cwgdj   1/1     Running   0          20s   10.244.1.104   node01   <none>           <none>


               
               
[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: test1
                operator: NotIn
                values:
                - a
#节点亲和性的硬策略，表示必须选择带有标签的值不是test1=a



[root@master01 k8s-yaml]# kubectl apply -f test12.yaml
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-7f4db4ff6b-4pg5w   1/1     Running   0          75s   10.244.2.145   node02   <none>           <none>
nginx1-7f4db4ff6b-dslv4   1/1     Running   0          76s   10.244.2.143   node02   <none>           <none>
nginx1-7f4db4ff6b-lmrjk   1/1     Running   0          76s   10.244.2.144   node02   <none>           <none>



试验：

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,test1=a,test2=b,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test1=a,test2=b

[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP       NODE     NOMINATED NODE   READINESS GATES
nginx1-7f4db4ff6b-99l2r   0/1     Pending   0          10s   <none>   <none>   <none>           <none>
nginx1-7f4db4ff6b-gjfkd   0/1     Pending   0          10s   <none>   <none>   <none>           <none>
nginx1-7f4db4ff6b-hsks5   0/1     Pending   0          10s   <none>   <none>   <none>           <none>



[root@master01 k8s-yaml]# kubectl label nodes node02 test1-
node/node02 labeled
[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,test1=a,test2=b,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b



[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-7f4db4ff6b-99l2r   1/1     Running   0          93s   10.244.2.146   node02   <none>           <none>
nginx1-7f4db4ff6b-gjfkd   1/1     Running   0          93s   10.244.2.147   node02   <none>           <none>
nginx1-7f4db4ff6b-hsks5   1/1     Running   0          93s   10.244.2.148   node02   <none>           <none>



条件满足，不需要强制重启，

[root@master01 k8s-yaml]# kubectl get pod -o wide即可

软策略：

单个权重

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            preference:
              matchExpressions:
              - key: test1
                operator: NotIn
                values:
                - a
#节点的亲和性的软策略，希望部署到不包含test1=a的标签节点。
[root@master01 k8s-yaml]# kubectl apply -f test12.yaml
[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                     READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-7b45d5b8f-hwxgj   1/1     Running   0          17s   10.244.2.149   node02   <none>           <none>
nginx1-7b45d5b8f-x24j4   1/1     Running   0          15s   10.244.2.150   node02   <none>           <none>
nginx1-7b45d5b8f-zbdvq   1/1     Running   0          14s   10.244.1.106   node01   <none>           <none>


##软限制，只能做一点点资源调度，不能强制



##podAffinity

[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        podAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            podAffinityTerm:
              labelSelector:
                matchExpressions:
                - key: app
                  operator: In
                  values:
                  - nginx1
              topologyKey: test3

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,memory=1000,test1=a,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b,xy102=98


[root@master01 k8s-yaml]# kubectl apply -f test13.yaml 


[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                      READY   STATUS    RESTARTS   AGE     IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j    1/1     Running   0          153m    10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56    1/1     Running   0          153m    10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj    1/1     Running   0          153m    10.244.2.152   node02   <none>           <none>
nginx2-6b6584f58f-2dr9n   1/1     Running   0          4m27s   10.244.1.120   node01   <none>           <none>
nginx2-6b6584f58f-7vntk   1/1     Running   0          4m27s   10.244.1.121   node01   <none>           <none>
nginx2-6b6584f58f-zbhv5   1/1     Running   0          4m27s   10.244.2.186   node02   <none>           <none>

多个权重

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            preference:
              matchExpressions:
              - key: test1
                operator: NotIn
                values:
                - a
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            preference:
              matchExpressions:
              - key: test2
                operator: NotIn
                values:
                - b
  #多个软策略以权重来进行分配，权重高的，优先级大


[root@master01 k8s-yaml]# kubectl apply -f test12.yaml
deployment.apps/nginx1 configured
[root@master01 k8s-yaml]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-698c85f477-475q8   1/1     Running   0          17s   10.244.1.107   node01   <none>           <none>
nginx1-698c85f477-jssbs   1/1     Running   0          15s   10.244.2.151   node02   <none>           <none>
nginx1-698c85f477-pc9m7   1/1     Running   0          14s   10.244.1.108   node01   <none>           <none>

  #多个软策略以权重来进行分配，权重高的，优先级大，weight=2>weight=1,使得在node01部署较多，不是绝对部署在node01

[root@master01 k8s-yaml]# vim test12.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx1
  name: nginx1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx1
  template:
    metadata:
      labels:
        app: nginx1
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            preference:
              matchExpressions:
              - key: test1
                operator: NotIn
                values:
                - a
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: test2
                operator: NotIn
                values:
                - b
#如果已经有硬策略，一般不需要声明软策略。

[root@master01 k8s-yaml]# kubectl apply -f test12.yaml
deployment.apps/nginx1 configured
[root@master01 k8s-yaml]# kubectl get pod -o wide

#匹配的pod的标签是app=nginx1，且节点上包含标签名是test2

[root@master01 k8s-yaml]# kubectl describe deployments.apps nginx1 



Labels:                 app=nginx1



[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,test1=a,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b




[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        podAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
          - labelSelector:
          #根据标签进行选择
               matchExpressions:
               - key: app
                 operator: In
                 values:
                 - nginx1
            topologyKey: test2
#匹配的pod的标签是app=nginx1，且节点上包含标签名是test2   





[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                     READY   STATUS    RESTARTS   AGE     IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j   1/1     Running   0          4m36s   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56   1/1     Running   0          4m36s   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj   1/1     Running   0          4m36s   10.244.2.152   node02   <none>           <none>
nginx2-8c5695d94-jx7sr   1/1     Running   0          4s      10.244.2.155   node02   <none>           <none>
nginx2-8c5695d94-qtcf6   1/1     Running   0          4s      10.244.2.153   node02   <none>           <none>
nginx2-8c5695d94-xwg6b   1/1     Running   0          4s      10.244.2.154   node02   <none>           <none>

3.2、反亲和性：

pod反亲和性 pod Anti-Affinity

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,test1=a,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b


[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
          - labelSelector:
          #根据标签进行选择
               matchExpressions:
               - key: app
                 operator: In
                 values:
                 - nginx1
            topologyKey: test2


##
[root@master01 k8s-yaml]# kubectl apply -f test13.yaml 
[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                     READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j   1/1     Running   0          13m   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56   1/1     Running   0          13m   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj   1/1     Running   0          13m   10.244.2.152   node02   <none>           <none>
nginx2-5784d9dcb-bnxcb   1/1     Running   0          72s   10.244.1.111   node01   <none>           <none>
nginx2-5784d9dcb-f6pvb   1/1     Running   0          70s   10.244.1.112   node01   <none>           <none>
nginx2-5784d9dcb-rpsjk   1/1     Running   0          69s   10.244.1.113   node01   <none>           <none>




#只能部署在pod的标签不是app=nginx1且节点的标签名不能有test2

#其实在pod的亲和性当中，起决定作用的是拓扑域的标签。




Events:
  Type     Reason            Age                From               Message
  ----     ------            ----               ----               -------
  Warning  FailedScheduling  22m (x2 over 22m)  default-scheduler  0/3 nodes are available: 3 node(s) didn't match pod affinity/anti-affinity, 3 node(s) didn't match pod anti-affinity rules.
  Normal   Scheduled         22m                default-scheduler  Successfully assigned default/nginx2-5bdc766cf7-sk69d to master01
  
  
  
#能够部署的原因在于调度器，调度器的核心是为了部署pod，只要能够部署，调度器有可能会忽略规则进行部署，既然有规则，调度器依然会部署
#资源不足的情况下，为了pod能够部署，调度器极有可能忽略所有的限制。

软策略：倾向性，尽可能的满足条件----更多的，尽量把资源调度到需要的节点

硬策略：必须要满足条件。-----特殊情况，节点故障，但是有业务要更新，强制性的把资源调度到指定的节点。

[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: test3
                operator: NotIn
                values:
                - b


[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j    1/1     Running   0          24m   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56    1/1     Running   0          24m   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj    1/1     Running   0          24m   10.244.2.152   node02   <none>           <none>
nginx2-5d77fdc494-2lxc6   1/1     Running   0          39s   10.244.2.157   node02   <none>           <none>
nginx2-5d77fdc494-7hswd   1/1     Running   0          38s   10.244.2.158   node02   <none>           <none>
nginx2-5d77fdc494-9ckqj   1/1     Running   0          40s   10.244.2.156   node02   <none>           <none>

Gt用法

[root@master01 k8s-yaml]# kubectl label nodes node01 memory=1000
node/node01 labeled
[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,memory=1000,test1=a,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b



[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: memory
                operator: Gt
                values:
                - "500"
[root@master01 k8s-yaml]# kubectl apply -f test13.yaml 

[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j    1/1     Running   0          43m   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56    1/1     Running   0          43m   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj    1/1     Running   0          43m   10.244.2.152   node02   <none>           <none>
nginx2-79b79dd568-gpjx5   1/1     Running   0          56s   10.244.1.116   node01   <none>           <none>
nginx2-79b79dd568-rhjs4   1/1     Running   0          58s   10.244.1.114   node01   <none>           <none>
nginx2-79b79dd568-trn54   1/1     Running   0          57s   10.244.1.115   node01   <none>           <none>

[root@master01 k8s-yaml]# vim test13.yaml 

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: memory
                operator: DoesNotExist
[root@master01 k8s-yaml]# kubectl apply -f test13.yaml 

[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j    1/1     Running   0          46m   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56    1/1     Running   0          46m   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj    1/1     Running   0          46m   10.244.2.152   node02   <none>           <none>
nginx2-585c559d74-ccqt6   1/1     Running   0          19s   10.244.2.159   node02   <none>           <none>
nginx2-585c559d74-k4vxw   1/1     Running   0          18s   10.244.2.160   node02   <none>           <none>
nginx2-585c559d74-zv6z5   1/1     Running   0          17s   10.244.2.161   node02   <none>           <none>

四、作业：

1、实现pod的探针：

就绪探针

tcpSocket

2、挂载，容器 /usr/share/nginx/html

节点 /opt/html

3、node的亲和性尽量部署在node01

4、pod的亲和性，尽量部署包含有app=nginx1的pod且目标签名是xy102的节点。

5、软策略选择标签名包含xy102, 值小于100。

[root@master01 k8s-yaml]# kubectl get nodes --show-labels
NAME       STATUS   ROLES                  AGE   VERSION    LABELS
master01   Ready    control-plane,master   8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=
node01     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node01,kubernetes.io/os=linux,memory=1000,test1=a,test3=b
node02     Ready    <none>                 8d    v1.20.15   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=node02,kubernetes.io/os=linux,test2=b,xy102=98


[root@master01 k8s-yaml]# vim test5.yaml 


apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx2
  name: nginx2
spec:
  replicas: 12
  selector:
    matchLabels:
      app: nginx2
  template:
    metadata:
      labels:
        app: nginx2
    spec:
      containers:
      - name: nginx
        image: nginx:1.22
        volumeMounts:
        - name: data-v
          mountPath: /usr/share/nginx/html
        readinessProbe:
          tcpSocket:
            port: 80
      volumes:
      - name: data-v
        hostPath:
          path: /opt/html
          type: DirectoryOrCreate
      affinity:
#选择亲和性的字段
        nodeAffinity:
#节点亲和性
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 3
            preference:
              matchExpressions:
              - key: test1
                operator: In
                values:
                - a
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            preference:
              matchExpressions:
              - key: xy102
                operator: Lt
                values:
                - "100"
        podAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
          - weight: 1
            podAffinityTerm:
              labelSelector:
                matchExpressions:
                - key: app
                  operator: In
                  values:
                  - nginx1
              topologyKey: xy102

[root@master01 k8s-yaml]# kubectl apply -f test5.yaml

[root@master01 k8s-yaml]# kubectl get pod -o wide 
NAME                      READY   STATUS    RESTARTS   AGE   IP             NODE     NOMINATED NODE   READINESS GATES
nginx1-654cb56c4-7t69j    1/1     Running   0          87m   10.244.1.110   node01   <none>           <none>
nginx1-654cb56c4-d8q56    1/1     Running   0          87m   10.244.1.109   node01   <none>           <none>
nginx1-654cb56c4-q89zj    1/1     Running   0          87m   10.244.2.152   node02   <none>           <none>
nginx2-6545dc89c4-4gdmz   1/1     Running   0          90s   10.244.1.119   node01   <none>           <none>
nginx2-6545dc89c4-675ch   1/1     Running   0          90s   10.244.2.176   node02   <none>           <none>
nginx2-6545dc89c4-8kz6b   1/1     Running   0          90s   10.244.2.184   node02   <none>           <none>
nginx2-6545dc89c4-b66r6   1/1     Running   0          90s   10.244.2.183   node02   <none>           <none>
nginx2-6545dc89c4-fc8bj   1/1     Running   0          90s   10.244.2.178   node02   <none>           <none>
nginx2-6545dc89c4-fp6nf   1/1     Running   0          90s   10.244.2.177   node02   <none>           <none>
nginx2-6545dc89c4-klxfc   1/1     Running   0          90s   10.244.2.180   node02   <none>           <none>
nginx2-6545dc89c4-llp4c   1/1     Running   0          90s   10.244.2.182   node02   <none>           <none>
nginx2-6545dc89c4-lw8l5   1/1     Running   0          90s   10.244.2.181   node02   <none>           <none>
nginx2-6545dc89c4-nrtwx   1/1     Running   0          90s   10.244.2.185   node02   <none>           <none>
nginx2-6545dc89c4-ssqrn   1/1     Running   0          90s   10.244.1.118   node01   <none>           <none>
nginx2-6545dc89c4-w49lx   1/1     Running   0          90s   10.244.2.179   node02   <none>           <none>