• 【云原生 | Kubernetes 实战】05、Pod高级实战：基于污点、容忍度、亲和性的多种调度策略（上）

    目录

    一、标签

    1.1 什么是标签？

    1.2 给pod资源打标签

    1.3 查看资源标签

    二、node 节点选择器

    2.1 nodeName：指定pod节点运行在哪个具体的node上

    2.2 nodeSelector：指定pod调度到具有特定标签的node节点上

    2.3 同时有 nodeName 和 nodeSelector 字段

    三、node 节点亲和性 

    3.1 使用 requiredDuringSchedulingIgnoredDuringExecution 硬亲和性

    3.2 使用 preferredDuringSchedulingIgnoredDuringExecution 软亲和性

    一、标签

    1.1 什么是标签？

            标签其实就一对 key/value ，被关联到对象上，比如Pod,标签的使用我们倾向于能够表示对象的特殊特点，就是一眼就看出了这个Pod是干什么的，标签可以用来划分特定的对象（比如版本，服务类型等），标签可以在创建一个对象的时候直接定义，也可以在后期随时修改，每一个对象可以拥有多个标签，但是，key值必须是唯一的。创建标签之后也可以方便我们对资源进行分组管理。如果对pod打标签，之后就可以使用标签来查看、删除指定的pod。

    在k8s中，大部分资源都可以打标签。

    1.2 给pod资源打标签

    1. #1. 创建pod资源
    2. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-first.yaml
    3.  
    4. #2. 查看默认空间下的所有 pods
    5. [root@k8s-master01 pod-yaml]# kubectl get pods 
    6. NAME          READY   STATUS    RESTARTS   AGE
    7. tomcat-test   1/1     Running   0          6s
    8.  
    9. #3. 查看默认空间下的所有pods 的标签
    10. [root@k8s-master01 pod-yaml]# kubectl get pods --show-labels 
    11. NAME          READY   STATUS    RESTARTS   AGE   LABELS
    12. tomcat-test   1/1     Running   0          38s   app=tomcat
    13.  
    14. #4. 对指定的pod资源 tomcat-test 打上标签
    15. [root@k8s-master01 pod-yaml]# kubectl label pods tomcat-test release=v1 
    16.  
    17. #5. 查看指定pods 的标签
    18. [root@k8s-master01 pod-yaml]# kubectl get pods tomcat-test --show-labels 
    19. NAME          READY   STATUS    RESTARTS   AGE     LABELS
    20. tomcat-test   1/1     Running   0          4m47s   app=tomcat,release=v1

    1.3 查看资源标签

    1. # 查看指定名称空间下的所有pods 的标签
    2. [root@k8s-master01 pod-yaml]# kubectl get pods -n kube-system --show-labels
    3.  
    4. # 查看默认名称空间下指定pod具有的所有标签
    5. [root@k8s-master01 pod-yaml]# kubectl get pods tomcat-test --show-labels
    6.  
    7. # 列出默认名称空间下标签key是release的pod，但不显示标签
    8. [root@k8s-master01 pod-yaml]# kubectl get pods -l release
    9. NAME          READY   STATUS    RESTARTS   AGE
    10. tomcat-test   1/1     Running   0          10m
    11.  
    12. # 列出默认名称空间下标签key是release、值是v1的pod，不显示标签
    13. [root@k8s-master01 pod-yaml]# kubectl get pods -l release=v1
    14. NAME          READY   STATUS    RESTARTS   AGE
    15. tomcat-test   1/1     Running   0          11m
    16.  
    17. # 列出默认名称空间下标签key是release的所有pod，并打印对应的标签值
    18. [root@k8s-master01 pod-yaml]# kubectl get pods -L release
    19. NAME          READY   STATUS    RESTARTS   AGE   RELEASE
    20. tomcat-test   1/1     Running   0          12m   v1
    21.  
    22. # 列出默认名称空间下多个标签（release,app）的所有pods，并打印对应的标签值
    23. [root@k8s-master01 pod-yaml]# kubectl get pods -L release,app
    24. NAME          READY   STATUS    RESTARTS   AGE   RELEASE   APP
    25. tomcat-test   1/1     Running   0          13m   v1        tomcat
    26.  
    27. # 查看所有名称空间下的所有pod的标签
    28. [root@k8s-master01 pod-yaml]# kubectl get pods --all-namespaces --show-labels

    二、node 节点选择器

            我们在创建pod资源的时候，pod会根据schduler进行调度，那么默认会调度到随机的一个工作节点，如果我们想要pod调度到指定节点或者调度到一些具有相同特点的node节点，怎么办呢？
    可以使用pod中的 nodeName 或者 nodeSelector 字段指定要调度到的node节点。

    2.1 nodeName：指定pod节点运行在哪个具体的node上

    1. #1. 在node1和node2上拉取 tomcat 和 busybox 镜像
    2. [root@k8s-node1 ~]# docker pull tomcat
    3. [root@k8s-node1 ~]# docker pull busybox
    4.  
    5. [root@k8s-node2 ~]# docker pull tomcat
    6. [root@k8s-node2 ~]# docker pull busybox
    7.  
    8. #2. 编写 yaml 文件
    9. [root@k8s-master01 pod-yaml]# vi pod-node.yaml
    10. apiVersion: v1
    11. kind: Pod
    12. metadata:
    13.   name: demo-pod
    14.   namespace: default
    15.   labels:
    16.     app: myapp
    17.     env: dev
    18. spec:
    19.   nodeName: k8s-node2    # 指定调度到 node2 节点
    20.   containers:
    21.   - name:  tomcat-pod
    22.     ports:
    23.     - containerPort: 8080
    24.     image: tomcat:latest
    25.     imagePullPolicy: IfNotPresent
    26.   - name: busybox
    27.     image: busybox:latest
    28.     command:             # busybox 容器指定 sh 编译器，-c 指定执行的命令
    29.     - "/bin/sh"
    30.     - "-c"
    31.     - "sleep 3600"
    32.  
    33. #3. 创建pod资源
    34. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-node.yaml 
    35. pod/demo-pod created
    36. [root@k8s-master01 pod-yaml]# kubectl get pods 
    37. NAME          READY   STATUS              RESTARTS   AGE
    38. demo-pod      0/2     ContainerCreating   0          7s
    39. tomcat-test   1/1     Running             0          32m
    40.  
    41. #4. 查看pods被调度到哪个节点了
    42. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    43. NAME          READY   STATUS    RESTARTS   AGE   IP               NODE        NOMINATED NODE   READINESS GATES
    44. demo-pod      2/2     Running   0          53s   10.244.169.140   k8s-node2   <none>           <none>
    45. tomcat-test   1/1     Running   0          33m   10.244.169.139   k8s-node2   <none>           <none>

    2.2 nodeSelector：指定pod调度到具有特定标签的node节点上

    1. #1. 查看k8s集群中所有node节点的标签
    2. [root@k8s-master01 pod-yaml]# kubectl get nodes --show-labels 
    3. NAME           STATUS   ROLES           AGE   VERSION   LABELS
    4. k8s-master01   Ready    control-plane   37d   v1.25.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-master01,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node.kubernetes.io/exclude-from-external-load-balancers=
    5. k8s-node1      Ready    work            37d   v1.25.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node1,kubernetes.io/os=linux,node-role.kubernetes.io/work=work
    6. k8s-node2      Ready    work            37d   v1.25.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node2,kubernetes.io/os=linux,node-role.kubernetes.io/work=work
    7.  
    8. #2. 编写 yaml 文件
    9. [root@k8s-master01 pod-yaml]# vi pod-1.yaml
    10. apiVersion: v1
    11. kind: Pod
    12. metadata:
    13.   name: demo-pod-1
    14.   namespace: default
    15.   labels:
    16.     app: myapp
    17.     env: dev
    18. spec:
    19.   nodeSelector:
    20.     disk: ceph    # 指定标签为 disk=ceph 的node节点
    21.   containers:
    22.   - name:  tomcat-pod-1
    23.     ports:
    24.     - containerPort: 8080
    25.     image: tomcat:latest
    26.     imagePullPolicy: IfNotPresent
    27.  
    28. #3. 创建 pod 资源
    29. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-1.yaml
    30.  
    31. #4. 查看pods状态。demo-pod-1 为等待状态
    32. [root@k8s-master01 pod-yaml]# kubectl get pods 
    33. NAME          READY   STATUS    RESTARTS   AGE
    34. demo-pod      2/2     Running   0          15m
    35. demo-pod-1    0/1     Pending   0          6s
    36. tomcat-test   1/1     Running   0          48m
    37.  
    38. #5. 查看pod详细信息
    39. [root@k8s-master01 pod-yaml]# kubectl describe pods demo-pod-1

    可以看到，demo-pod-1 资源调度失败，三个节点没有可用的。因为找不到指定标签 disk: ceph 的node：

    解决办法：

    1. #1. 给node1节点打标签，打个具有disk=ceph的标签
    2. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node1 disk=ceph
    3.  
    4. #2. 再次查看pods状态
    5. [root@k8s-master01 pod-yaml]# kubectl get pods 
    6. NAME          READY   STATUS    RESTARTS   AGE
    7. demo-pod      2/2     Running   0          25m
    8. demo-pod-1    1/1     Running   0          10m
    9. tomcat-test   1/1     Running   0          58m
    10.  
    11. #3. 查看pods 调度信息
    12. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    13. NAME          READY   STATUS    RESTARTS   AGE   IP               NODE        NOMINATED NODE   READINESS GATES
    14. demo-pod      2/2     Running   0          27m   10.244.169.140   k8s-node2   <none>           <none>
    15. demo-pod-1    1/1     Running   0          11m   10.244.36.75     k8s-node1   <none>           <none>
    16. tomcat-test   1/1     Running   0          60m   10.244.169.139   k8s-node2   <none>           <none>
    17.  
    18. #4. 再次查看 node 节点的标签
    19. [root@k8s-master01 pod-yaml]# kubectl get nodes --show-labels

    做完上面实验，需要把default名称空间下的pod全都删除，kubectl delete pods pod名字：

    1. [root@k8s-master01 ~]# kubectl delete pods demo-pod demo-pod-1 tomcat-test 
    2. pod "demo-pod" deleted
    3. pod "demo-pod-1" deleted
    4. pod "tomcat-test" deleted
    5. [root@k8s-master01 ~]# kubectl get pods 
    6. No resources found in default namespace.
    7.  
    8. # 删除node1节点打的标签 disk=ceph 。把等号换成 '-' 表示删除
    9. [root@k8s-master01 ~]# kubectl label nodes k8s-node1 disk-
    10.  
    11. # 查看是否删除成功
    12. [root@k8s-master01 ~]# kubectl get nodes k8s-node1 --show-labels

    2.3 同时有 nodeName 和 nodeSelector 字段

    1. [root@k8s-master01 pod-yaml]# vi pod-1.yaml
    2. apiVersion: v1
    3. kind: Pod
    4. metadata:
    5.   name: demo-pod-1
    6.   namespace: default
    7.   labels:
    8.     app: myapp
    9.     env: dev
    10. spec:
    11.   nodeName: k8s-node2
    12.   nodeSelector:
    13.     disk: ceph
    14.   containers:
    15.   - name:  tomcat-pod-1
    16.     ports:
    17.     - containerPort: 8080
    18.     image: tomcat:latest
    19.     imagePullPolicy: IfNotPresent
    20.  
    21. # 创建pod资源
    22. [root@k8s-master01 pod-yaml]# kubectl get pods 
    23. NAME         READY   STATUS         RESTARTS   AGE
    24. demo-pod-1   0/1     NodeAffinity   0          7s
    25. [root@k8s-master01 pod-yaml]# kubectl describe pods demo-pod-1

    创建失败了，不能正常调度，报错信息如下：

            结论：同一个yaml文件里定义pod资源，如果同时定义了nodeName和NodeSelector，那么条件必须都满足才可以，有一个不满足都会调度失败。

    1. # 我给node2打上标签 disk=ceph
    2. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node2 disk=ceph
    3.  
    4. [root@k8s-master01 pod-yaml]# kubectl delete pods demo-pod-1
    5.  
    6. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-1.yaml
    7.  
    8. # 调度成功
    9. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    10. NAME         READY   STATUS    RESTARTS   AGE   IP               NODE        NOMINATED NODE   READINESS GATES
    11. demo-pod-1   1/1     Running   0          5s    10.244.169.143   k8s-node2   <none>           <none>

    结论：如果同时定义了nodeName和NodeSelector，如果只满足NodeSelector则可以调度成功。

    三、node 节点亲和性 

    node节点亲和性调度：nodeAffinity

    官方文档：将 Pod 指派给节点 | Kubernetes

    1. # 查看帮助命令。一层一层的找
    2. [root@k8s-master01 pod-yaml]# kubectl explain pods.spec.affinity
    3. KIND:     Pod
    4. VERSION:  v1
    5.  
    6. RESOURCE: affinity <Object>
    7.  
    8. DESCRIPTION:
    9.      If specified, the pod's scheduling constraints
    10.  
    11.      Affinity is a group of affinity scheduling rules.
    12.  
    13. FIELDS:
    14.    nodeAffinity	<Object>
    15.      Describes node affinity scheduling rules for the pod.
    16.  
    17.    podAffinity	<Object>
    18. ······
    19.  
    20.    podAntiAffinity	<Object>
    21. ······
    22.  
    23. [root@k8s-master01 pod-yaml]# kubectl explain pods.spec.affinity.nodeAffinity
    24. KIND:     Pod
    25. VERSION:  v1
    26.  
    27. RESOURCE: nodeAffinity <Object>
    28.  
    29. DESCRIPTION:
    30.      Describes node affinity scheduling rules for the pod.
    31.  
    32.      Node affinity is a group of node affinity scheduling rules.
    33.  
    34. FIELDS:
    35.    preferredDuringSchedulingIgnoredDuringExecution	<[]Object>
    36.      The scheduler will prefer to schedule pods to nodes that satisfy the
    37.      affinity expressions specified by this field, but it may choose a node that
    38.      violates one or more of the expressions. The node that is most preferred is
    39.      the one with the greatest sum of weights, i.e. for each node that meets all
    40.      of the scheduling requirements (resource request, requiredDuringScheduling
    41.      affinity expressions, etc.), compute a sum by iterating through the
    42.      elements of this field and adding "weight" to the sum if the node matches
    43.      the corresponding matchExpressions; the node(s) with the highest sum are
    44.      the most preferred.
    45.  
    46.    requiredDuringSchedulingIgnoredDuringExecution	<Object>
    47.      If the affinity requirements specified by this field are not met at
    48.      scheduling time, the pod will not be scheduled onto the node. If the
    49.      affinity requirements specified by this field cease to be met at some point
    50.      during pod execution (e.g. due to an update), the system may or may not try
    51.      to eventually evict the pod from its node.
    52.  
    53. # prefered：表示有节点尽量满足这个位置定义的亲和性，这不是一个必须的条件，软亲和性
    54. # require：表示必须有节点满足这个位置定义的亲和性，这是个硬性条件，硬亲和性
    55.  
    56. [root@k8s-master01 pod-yaml]# kubectl explain pods.spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution
    57. KIND:     Pod
    58. VERSION:  v1
    59.  
    60. RESOURCE: requiredDuringSchedulingIgnoredDuringExecution <Object>
    61.  
    62. DESCRIPTION:
    63.      If the affinity requirements specified by this field are not met at
    64.      scheduling time, the pod will not be scheduled onto the node. If the
    65.      affinity requirements specified by this field cease to be met at some point
    66.      during pod execution (e.g. due to an update), the system may or may not try
    67.      to eventually evict the pod from its node.
    68.  
    69.      A node selector represents the union of the results of one or more label
    70.      queries over a set of nodes; that is, it represents the OR of the selectors
    71.      represented by the node selector terms.
    72.  
    73. FIELDS:
    74.    nodeSelectorTerms	<[]Object> -required-
    75.      Required. A list of node selector terms. The terms are ORed.
    76.  
    77. [root@k8s-master01 pod-yaml]# kubectl explain pods.spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms
    78. KIND:     Pod
    79. VERSION:  v1
    80.  
    81. RESOURCE: nodeSelectorTerms <[]Object>
    82.  
    83. DESCRIPTION:
    84.      Required. A list of node selector terms. The terms are ORed.
    85.  
    86.      A null or empty node selector term matches no objects. The requirements of
    87.      them are ANDed. The TopologySelectorTerm type implements a subset of the
    88.      NodeSelectorTerm.
    89.  
    90. FIELDS:
    91.    # 匹配表达式的
    92.    matchExpressions	<[]Object>
    93.      A list of node selector requirements by node's labels.
    94.   
    95.    # 匹配字段的
    96.    matchFields	<[]Object>
    97.      A list of node selector requirements by node's fields.
    98.  
    99. [root@k8s-master01 pod-yaml]# kubectl explain pods.spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms.matchFields
    100. KIND:     Pod
    101. VERSION:  v1
    102.  
    103. RESOURCE: matchFields <[]Object>
    104.  
    105. DESCRIPTION:
    106.      A list of node selector requirements by node's fields.
    107.  
    108.      A node selector requirement is a selector that contains values, a key, and
    109.      an operator that relates the key and values.
    110.  
    111. FIELDS:
    112.    # 检查label标签
    113.    key	<string> -required-
    114.      The label key that the selector applies to.
    115.  
    116.    # 做等值选则还是不等值选则
    117.    operator	<string> -required-
    118.      Represents a key's relationship to a set of values. Valid operators are In,
    119.      NotIn, Exists, DoesNotExist. Gt, and Lt.
    120.  
    121.      Possible enum values:
    122.      - `"DoesNotExist"`
    123.      - `"Exists"`
    124.      - `"Gt"`
    125.      - `"In"`
    126.      - `"Lt"`
    127.      - `"NotIn"`
    128.  
    129.    # 给定值
    130.    values	<[]string>
    131.      An array of string values. If the operator is In or NotIn, the values array
    132.      must be non-empty. If the operator is Exists or DoesNotExist, the values
    133.      array must be empty. If the operator is Gt or Lt, the values array must
    134.      have a single element, which will be interpreted as an integer. This array
    135.      is replaced during a strategic merge patch.

    3.1 使用 requiredDuringSchedulingIgnoredDuringExecution 硬亲和性

    1. # 检查当前节点中有任意一个节点拥有zone标签的值是foo或者bar，就可以把pod调度到这个node节点的foo或者bar标签上的节点上
    2. [root@k8s-master01 pod-yaml]# vim pod-nodeaffinity-demo.yaml
    3. apiVersion: v1
    4. kind: Pod
    5. metadata:
    6.   name:  pod-node-affinity-demo
    7.   namespace: default
    8.   labels:
    9.     app: myapp
    10.     tier: frontend
    11. spec:
    12.   affinity:
    13.     nodeAffinity:
    14.       requiredDuringSchedulingIgnoredDuringExecution:
    15.         nodeSelectorTerms:
    16.         - matchExpressions:
    17.           - key: zone
    18.             operator: In
    19.             values:
    20.             - foo
    21.             - bar
    22.   containers:
    23.   - name: myapp
    24.     image: nginx:latest
    25.     imagePullPolicy: IfNotPresent
    26.  
    27. [root@k8s-master01 pod-yaml]# kubectl get pods 
    28. NAME                     READY   STATUS    RESTARTS   AGE
    29. demo-pod-1               1/1     Running   0          38m
    30. pod-node-affinity-demo   0/1     Pending   0          8s
    31.  
    32. [root@k8s-master01 pod-yaml]# kubectl describe pods pod-node-affinity-demo

            status 的状态是pending，上面说明没有完成调度，因为没有一个拥有zone的标签的值是foo或者bar，而且使用的是硬亲和性，必须满足条件才能完成调度：

    1. # 给 node1 打上标签 zone=foo
    2. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node1 zone=foo
    3. node/k8s-node1 labeled
    4. 您在 /var/spool/mail/root 中有新邮件
    5.  
    6. [root@k8s-master01 pod-yaml]# kubectl get pods 
    7. NAME                     READY   STATUS    RESTARTS   AGE
    8. demo-pod-1               1/1     Running   0          43m
    9. pod-node-affinity-demo   1/1     Running   0          4m47s
    10.  
    11. # 再次查看，调度成功
    12. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    13. NAME                     READY   STATUS    RESTARTS   AGE     IP               NODE        NOMINATED NODE   READINESS GATES
    14. demo-pod-1               1/1     Running   0          43m     10.244.169.143   k8s-node2   <none>           <none>
    15. pod-node-affinity-demo   1/1     Running   0          4m52s   10.244.36.79     k8s-node1   <none>           <none>

    3.2 使用 preferredDuringSchedulingIgnoredDuringExecution 软亲和性

    1. [root@k8s-master01 pod-yaml]# vim pod-nodeaffinity-demo-2.yaml
    2. apiVersion: v1
    3. kind: Pod
    4. metadata:
    5.   name: pod-node-affinity-demo-2
    6.   namespace: default
    7.   labels:
    8.     app: myapp02
    9.     tier: frontend
    10. spec:
    11.   containers:
    12.   - name: myapp02
    13.     image: nginx:latest
    14.     imagePullPolicy: IfNotPresent
    15.   affinity:
    16.     nodeAffinity:
    17.       preferredDuringSchedulingIgnoredDuringExecution:
    18.       - preference:
    19.           matchExpressions: 
    20.           - key: zone1
    21.             operator: In
    22.             values:
    23.             - foo1
    24.             - bar1
    25.         weight: 10
    26.       - preference:
    27.           matchExpressions:
    28.           - key: zone2
    29.             operator: In
    30.             values:
    31.             - foo2
    32.             - bar2
    33.         weight: 20
    34.  
    35. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-nodeaffinity-demo-2.yaml
    36.  
    37. [root@k8s-master01 pod-yaml]# kubectl get pods
    38. NAME                       READY   STATUS    RESTARTS        AGE
    39. demo-pod-1                 1/1     Running   1 (9m39s ago)   24h
    40. pod-node-affinity-demo     1/1     Running   1 (9m42s ago)   23h
    41. pod-node-affinity-demo-2   1/1     Running   0               4s
    42. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    43. NAME                       READY   STATUS    RESTARTS        AGE   IP               NODE        NOMINATED NODE   READINESS GATES
    44. demo-pod-1                 1/1     Running   1 (9m52s ago)   24h   10.244.169.144   k8s-node2   <none>           <none>
    45. pod-node-affinity-demo     1/1     Running   1 (9m55s ago)   23h   10.244.36.81     k8s-node1   <none>           <none>
    46. pod-node-affinity-demo-2   1/1     Running   0               17s   10.244.169.145   k8s-node2   <none>           <none>

    上面说明软亲和性是可以运行这个pod的，尽管没有节点被定义标签 zone1、zone2，pod依然会被调度（随机）。

            下面来测试 weight 权重。weight 是相对权重，权重值越高，pod被调度到指定的node节点几率就越大。

    1. # 给node1和node2打上标签：
    2. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node1 zone1=foo1
    3.  
    4. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node2 zone2=foo2
    5.  
    6. # 删除原来的pod：
    7. [root@k8s-master01 pod-yaml]# kubectl delete pods pod-node-affinity-demo-2
    8.  
    9. # 修改下权重，node1 权重为40
    10. [root@k8s-master01 pod-yaml]# vim pod-nodeaffinity-demo-2.yaml
    11.       preferredDuringSchedulingIgnoredDuringExecution:
    12.       - preference:
    13.           matchExpressions: 
    14.           - key: zone1
    15.             operator: In
    16.             values:
    17.             - foo1
    18.             - bar1
    19.         weight: 40
    20.       - preference:
    21.           matchExpressions:
    22.           - key: zone2
    23.             operator: In
    24.             values:
    25.             - foo2
    26.             - bar2
    27.         weight: 20
    28.  
    29. # 被调度到node1了。前面是被随机调度到node2的
    30. [root@k8s-master01 pod-yaml]# kubectl apply -f pod-nodeaffinity-demo-2.yaml 
    31. pod/pod-node-affinity-demo-2 created
    32. [root@k8s-master01 pod-yaml]# kubectl get pods -o wide 
    33. NAME                       READY   STATUS    RESTARTS      AGE   IP               NODE        NOMINATED NODE   READINESS GATES
    34. demo-pod-1                 1/1     Running   1 (20m ago)   24h   10.244.169.144   k8s-node2   <none>           <none>
    35. pod-node-affinity-demo     1/1     Running   1 (20m ago)   23h   10.244.36.81     k8s-node1   <none>           <none>
    36. pod-node-affinity-demo-2   1/1     Running   0             4s    10.244.36.83     k8s-node1   <none>           <none>

            结论：pod在定义node节点亲和性的时候，如果node1 和 node2 都满足相同的条件（如标签相同、weight 相同），都可以调度pod。但是node2具有的标签是zone2=foo2，pod在匹配zone2=foo2的权重高，那么pod就会优先调度到onode2上。

    1. # 把前面做实验的pod资源删除。为后面实验做准备
    2. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node1 zone1-
    3. node/k8s-node1 unlabeled
    4. 您在 /var/spool/mail/root 中有新邮件
    5. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node1 zone-
    6. node/k8s-node1 unlabeled
    7. [root@k8s-master01 pod-yaml]# kubectl label nodes k8s-node2 zone2-
    8. node/k8s-node2 unlabeled
    9. [root@k8s-master01 pod-yaml]# kubectl delete pods demo-pod-1 pod-node-affinity-demo pod-node-affinity-demo-2

    上一篇文章：【云原生 | Kubernetes 实战】04、k8s 名称空间和资源配额_Stars.Sky的博客-CSDN博客

    下一篇文章：Pod高级实战：基于污点、容忍度、亲和性的多种调度策略（下）

  • 相关阅读:
    Nvidia Jetson平台 JP4.2.1版本GStreamer视频采集异常问题调试记录
    MySQL：至少参与xxx参与的全部事件
    前端必备工具
    音视频（1） - FFmpeg4.3.4编译
    iOS打基础之Block二三事
    正则表达式简介
    解读GaussDB(for MySQL)灵活多维的二级分区表策略
    GBase 8a MPP集群管理之虚拟集群镜像表
    Reentrantlock简介及使用场景
    UI设计师岗位的基本职责八篇
  • 原文地址：https://blog.csdn.net/weixin_46560589/article/details/128086399