k8s运行容器之deployment（三）

靠谱运维
靠谱运维
靠谱运维
211
文章
18
评论
2020年3月25日12:50:05容器技术评论3,143

deployment

我们已经知道k8s是通过各种controller来管理pod的生命周期。为了满足不同业务场景，k8s开发了Deployment、ReplicaSet、DaemonSet、StatefuleSet、Job 等多种 Controller。我们首先学习最常用的 Deployment。文章源自靠谱运维-https://www.ixdba.net/archives/1363

运行一个deployment文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl run httpd-ken1--generator=run-pod/v1 --image=httpd --replicas=2

下面详细分析 Kubernetes 都做了些什么工作。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get deployment
NAME        READY   UP-TO-DATE   AVAILABLE   AGE
httpd-ken   2/2     2            2           35m

kubectl get deplouyment命令可以查看 httpd-ken 的状态，输出显示两个副本正常运行。文章源自靠谱运维-https://www.ixdba.net/archives/1363

接下来我们用 kubectl describe deployment 了解更详细的信息。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl  describe deployment httpd-ken
Name:                   httpd-ken
Namespace:              default
CreationTimestamp:      Tue, 29 Jan 2019 15:27:40 +0800
Labels:                 run=httpd-ken
Annotations:            deployment.kubernetes.io/revision: 1
Selector:               run=httpd-ken
Replicas:               2 desired | 2 updated | 2 total | 2 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        0
RollingUpdateStrategy:  25% max unavailable, 25% max surge
Pod Template:
  Labels:  run=httpd-ken
  Containers:
   httpd-ken:
    Image:        httpd
    Port:         <none>
    Host Port:    <none>
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      True    MinimumReplicasAvailable
  Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   httpd-ken-5c949b96f (2/2 replicas created)
Events:
  Type    Reason             Age   From                   Message
  ----    ------             ----  ----                   -------
  Normal  ScalingReplicaSet  18m   deployment-controller  Scaled up replica set httpd-ken-5c949b96f to 2

大部分内容都是自解释的，我们重点看最下面部分。这里告诉我们创建了一个 ReplicaSet httpd-ken-5c949b96，Events 是 Deployment 的日志，记录了 ReplicaSet 的启动过程。文章源自靠谱运维-https://www.ixdba.net/archives/1363

通过上面的分析，也验证了 Deployment 通过 ReplicaSet 来管理 Pod 的事实。接着我们将注意力切换到 httpd-ken-5c949b96，执行 kubectl describe replicaset：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get replicaset
NAME                  DESIRED   CURRENT   READY   AGE
httpd-ken-5c949b96f   2         2         2       20m

两个副本已经就绪，用 kubectl describe replicaset 查看详细信息：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl describe replicaset
Name:           httpd-ken-5c949b96f
Namespace:      default
Selector:       pod-template-hash=5c949b96f,run=httpd-ken
Labels:         pod-template-hash=5c949b96f
                run=httpd-ken
Annotations:    deployment.kubernetes.io/desired-replicas: 2
                deployment.kubernetes.io/max-replicas: 3
                deployment.kubernetes.io/revision: 1
Controlled By:  Deployment/httpd-ken
Replicas:       2 current / 2 desired
Pods Status:    2 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  pod-template-hash=5c949b96f
           run=httpd-ken
  Containers:
   httpd-ken:
    Image:        httpd
    Port:         <none>
    Host Port:    <none>
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:
  Type    Reason            Age   From                   Message
  ----    ------            ----  ----                   -------
  Normal  SuccessfulCreate  20m   replicaset-controller  Created pod: httpd-ken-5c949b96f-twdsd
  Normal  SuccessfulCreate  20m   replicaset-controller  Created pod: httpd-ken-5c949b96f-9cd52

Controlled By 指明此 ReplicaSet 是由 Deployment httpd-ken 创建。Events 记录了两个副本 Pod 的创建。接着我们来看 Pod，执行 kubectl get pod：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get pod
NAME                        READY   STATUS    RESTARTS   AGE
httpd-ken-5c949b96f-9cd52   1/1     Running   0          22m
httpd-ken-5c949b96f-twdsd   1/1     Running   0          22m

两个副本 Pod 都处于 Running 状态，用 kubectl describe pod 查看更详细的信息：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

root@ken ~]# kubectl describe pod
Name:               httpd-ken-5c949b96f-9cd52
Namespace:          default
Priority:           0
PriorityClassName:  <none>
Node:               host1/172.20.10.7
Start Time:         Tue, 29 Jan 2019 15:46:45 +0800
Labels:             pod-template-hash=5c949b96f
                    run=httpd-ken
Annotations:        <none>
Status:             Running
IP:                 10.244.1.3
Controlled By:      ReplicaSet/httpd-ken-5c949b96f
Containers:
  httpd-ken:
    Container ID:   docker://e59bda9941a16f20027c89a0d8fa8e17797b517f6f5461e905c0d29b57369dde
    Image:          httpd
    Image ID:       docker-pullable://httpd@sha256:44daa8e932a32ab6e50636d769ca9a60ad412124653707e5ed59c0209c72f9b3
    Port:           <none>
    Host Port:      <none>
    State:          Running
      Started:      Tue, 29 Jan 2019 15:47:10 +0800
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-vb7lm (ro)
Conditions:
  Type              Status
  Initialized       True 
  Ready             True 
  ContainersReady   True 
  PodScheduled      True 
Volumes:
  default-token-vb7lm:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-vb7lm
    Optional:    false
QoS Class:       BestEffort
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type    Reason     Age   From               Message
  ----    ------     ----  ----               -------
  Normal  Scheduled  23m   default-scheduler  Successfully assigned default/httpd-ken-5c949b96f-9cd52 to host1
  Normal  Pulling    23m   kubelet, host1     pulling image "httpd"
  Normal  Pulled     22m   kubelet, host1     Successfully pulled image "httpd"
  Normal  Created    22m   kubelet, host1     Created container
  Normal  Started    22m   kubelet, host1     Started container

Controlled By 指明此 Pod 是由 ReplicaSet  httpd-ken-5c949b96f创建。Events 记录了 Pod 的启动过程。如果操作失败（比如 image 不存在），也能在这里查看到原因。文章源自靠谱运维-https://www.ixdba.net/archives/1363

总结一下这个过程：文章源自靠谱运维-https://www.ixdba.net/archives/1363

用户通过 kubectl 创建 Deployment。文章源自靠谱运维-https://www.ixdba.net/archives/1363

Deployment 创建 ReplicaSet。文章源自靠谱运维-https://www.ixdba.net/archives/1363

ReplicaSet 创建 Pod文章源自靠谱运维-https://www.ixdba.net/archives/1363

也可以看出，对象的命名方式是：子对象的名字 = 父对象名字 + 随机字符串或数字。文章源自靠谱运维-https://www.ixdba.net/archives/1363

命令vs配置文件

k8s支持两种创建资源的方式：文章源自靠谱运维-https://www.ixdba.net/archives/1363

1.用 kubectl 命令直接创建，比如：文章源自靠谱运维-https://www.ixdba.net/archives/1363

kubectl run nginx-deployment –image=nginx:1.7.9 –replicas=2文章源自靠谱运维-https://www.ixdba.net/archives/1363

在命令行中通过参数指定资源的属性。文章源自靠谱运维-https://www.ixdba.net/archives/1363

2. 通过配置文件和 kubectl apply 创建，要完成前面同样的工作，可执行命令：文章源自靠谱运维-https://www.ixdba.net/archives/1363

kubectl apply -f nginx.yml文章源自靠谱运维-https://www.ixdba.net/archives/1363

nginx.yml 的内容为：文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

资源的属性写在配置文件中，文件格式为 YAML。文章源自靠谱运维-https://www.ixdba.net/archives/1363

下面对这两种方式进行比较。文章源自靠谱运维-https://www.ixdba.net/archives/1363

基于命令的方式：文章源自靠谱运维-https://www.ixdba.net/archives/1363

简单直观快捷，上手快。文章源自靠谱运维-https://www.ixdba.net/archives/1363

适合临时测试或实验。文章源自靠谱运维-https://www.ixdba.net/archives/1363

基于配置文件的方式：文章源自靠谱运维-https://www.ixdba.net/archives/1363

配置文件描述了 What，即应用最终要达到的状态。文章源自靠谱运维-https://www.ixdba.net/archives/1363

配置文件提供了创建资源的模板，能够重复部署。文章源自靠谱运维-https://www.ixdba.net/archives/1363

可以像管理代码一样管理部署。文章源自靠谱运维-https://www.ixdba.net/archives/1363

适合正式的、跨环境的、规模化部署。文章源自靠谱运维-https://www.ixdba.net/archives/1363

这种方式要求熟悉配置文件的语法，有一定难度。文章源自靠谱运维-https://www.ixdba.net/archives/1363

后面我们都将采用配置文件的方式，大家需要尽快熟悉和掌握。文章源自靠谱运维-https://www.ixdba.net/archives/1363

kubectl apply 不但能够创建 Kubernetes 资源，也能对资源进行更新，非常方便。不过 Kubernets 还提供了几个类似的命令，例如 kubectl create、kubectl replace、kubectl edit 和 kubectl patch。文章源自靠谱运维-https://www.ixdba.net/archives/1363

为避免造成不必要的困扰，我们会尽量只使用 kubectl apply，文章源自靠谱运维-https://www.ixdba.net/archives/1363

此命令已经能够应对超过 90% 的场景，事半功倍。文章源自靠谱运维-https://www.ixdba.net/archives/1363

deployment yaml

既然要用 YAML 配置文件部署应用，现在就很有必要了解一下 Deployment 的配置格式，其他 Controller（比如 DaemonSet）非常类似。文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

① apiVersion 是当前配置格式的版本。文章源自靠谱运维-https://www.ixdba.net/archives/1363

先执行kubectl api-resources找到所有的资源文章源自靠谱运维-https://www.ixdba.net/archives/1363

在执行命令 kubectl explain deploy即可获取到版本和类型信息文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

② kind 是要创建的资源类型，这里是 Deployment。文章源自靠谱运维-https://www.ixdba.net/archives/1363

③ metadata 是该资源的元数据，name 是必需的元数据项。文章源自靠谱运维-https://www.ixdba.net/archives/1363

④ spec 部分是该 Deployment 的规格说明。文章源自靠谱运维-https://www.ixdba.net/archives/1363

⑤ replicas 指明副本数量，默认为 1。文章源自靠谱运维-https://www.ixdba.net/archives/1363

⑥ template 定义 Pod 的模板，这是配置文件的重要部分。文章源自靠谱运维-https://www.ixdba.net/archives/1363

⑦ metadata 定义 Pod 的元数据，至少要定义一个 label。label 的 key 和 value 可以任意指定。文章源自靠谱运维-https://www.ixdba.net/archives/1363

⑧ spec 描述 Pod 的规格，此部分定义 Pod 中每一个容器的属性，name 和 image 是必需的。文章源自靠谱运维-https://www.ixdba.net/archives/1363

注意：最后name前面加个横线是因为container后面是一个列表文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

此 nginx.yml 是一个最简单的 Deployment 配置文件，后面我们学习 Kubernetes 各项功能时会逐步丰富这个文件。文章源自靠谱运维-https://www.ixdba.net/archives/1363

执行 kubectl apply -f nginx.yml：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
deployment.extensions/nginx-deployment created

查看nginx-deployment各种资源文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get deployment
NAME               READY   UP-TO-DATE   AVAILABLE   AGE
httpd-ken          2/2     2            2           73m
nginx-deployment   2/2     2            2           107s
[root@ken ~]# kubectl get replicaset
NAME                          DESIRED   CURRENT   READY   AGE
httpd-ken-5c949b96f           2         2         2       54m
nginx-deployment-65998d8886   2         2         2       111s
[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE    IP           NODE    NOMINATED NODE   READINESS GATES
httpd-ken-5c949b96f-9cd52           1/1     Running   0          54m    10.244.1.3   host1   <none>           <none>
httpd-ken-5c949b96f-twdsd           1/1     Running   0          54m    10.244.2.3   host2   <none>           <none>
nginx-deployment-65998d8886-9qrrv   1/1     Running   0          2m4s   10.244.2.4   host2   <none>           <none>
nginx-deployment-65998d8886-vnbgt   1/1     Running   0          2m4s   10.244.1.4   host1   <none>           <none>

Deployment、ReplicaSet、Pod 都已经就绪。如果要删除这些资源，执行 kubectl delete deployment nginx-deployment 或者 kubectl delete -f nginx.yml （编写的nginx.yml文件不会被删除）。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl delete -f nginx.yml
deployment.extensions "nginx-deployment" deleted

Scale Up/Down

伸缩（Scale Up/Down）是指在线增加或减少 Pod 的副本数。文章源自靠谱运维-https://www.ixdba.net/archives/1363

Deployment nginx-deployment 初始是两个副本。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE   IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-65998d8886-5b5rc   1/1     Running   0          84s   10.244.1.5   host1   <none>           <none>
nginx-deployment-65998d8886-tnpcx   1/1     Running   0          84s   10.244.2.5   host2   <none>           <none>

k8s-node1 和 k8s-node2 上各跑了一个副本。现在修改 nginx.yml，将副本改成 5 个。文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

再次执行kubectl apply文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
deployment.extensions/nginx-deployment configured

查看pod文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE     IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-65998d8886-4hfgp   1/1     Running   0          3m      10.244.1.7   host1   <none>           <none>
nginx-deployment-65998d8886-5b5rc   1/1     Running   0          5m48s   10.244.1.5   host1   <none>           <none>
nginx-deployment-65998d8886-btrsq   1/1     Running   0          3m      10.244.2.6   host2   <none>           <none>
nginx-deployment-65998d8886-tnpcx   1/1     Running   0          5m48s   10.244.2.5   host2   <none>           <none>
nginx-deployment-65998d8886-x4pbd   1/1     Running   0          3m      10.244.1.6   host1   <none>           <none>

三个新副本被创建并调度到 k8s-node1 和 k8s-node2 上。文章源自靠谱运维-https://www.ixdba.net/archives/1363

接下来修改配置文件，将副本数减少为 3 个，重新执行 kubectl apply：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
deployment.extensions/nginx-deployment configured
[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE     IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-65998d8886-5b5rc   1/1     Running   0          7m6s    10.244.1.5   host1   <none>           <none>
nginx-deployment-65998d8886-btrsq   1/1     Running   0          4m18s   10.244.2.6   host2   <none>           <none>
nginx-deployment-65998d8886-tnpcx   1/1     Running   0          7m6s    10.244.2.5   host2   <none>           <none>

可以看到两个副本被删除，最终保留了 3 个副本。文章源自靠谱运维-https://www.ixdba.net/archives/1363

模拟故障

上面我们有 3 个 nginx 副本分别运行在 k8s-node1 和 k8s-node2 上。现在模拟 k8s-node2 故障，关闭该节点（poweroff）。文章源自靠谱运维-https://www.ixdba.net/archives/1363

首先查看节点文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get node
NAME    STATUS     ROLES    AGE     VERSION
host1   Ready      <none>   5h25m   v1.13.2
host2   NotReady   <none>   5h43m   v1.13.2
ken     Ready      master   6h18m   v1.13.2

发现host2状态为NotReady文章源自靠谱运维-https://www.ixdba.net/archives/1363

等待一段时间，Kubernetes 会检查到 k8s-node2 不可用，将 k8s-node2 上的 Pod 标记为  Terminating状态，并在 k8s-node1 上新创建两个 Pod，维持总副本数为 3。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS        RESTARTS   AGE   IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-65998d8886-5b5rc   1/1     Running       0          16m   10.244.1.5   host1   <none>           <none>
nginx-deployment-65998d8886-8647d   1/1     Running       0          79s   10.244.1.8   host1   <none>           <none>
nginx-deployment-65998d8886-btrsq   1/1     Terminating   0          13m   10.244.2.6   host2   <none>           <none>
nginx-deployment-65998d8886-qp6jj   1/1     Running       0          79s   10.244.1.9   host1   <none>           <none>
nginx-deployment-65998d8886-tnpcx   1/1     Terminating   0          16m   10.244.2.5   host2   <none>           <none>

当 k8s-node2 恢复后， Terminating的 Pod 会被删除，不过已经运行的 Pod 不会重新调度回 k8s-node2。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get node
NAME    STATUS   ROLES    AGE     VERSION
host1   Ready    <none>   5h33m   v1.13.2
host2   Ready    <none>   5h51m   v1.13.2
ken     Ready    master   6h26m   v1.13.2
[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE     IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-65998d8886-5b5rc   1/1     Running   0          19m     10.244.1.5   host1   <none>           <none>
nginx-deployment-65998d8886-8647d   1/1     Running   0          4m33s   10.244.1.8   host1   <none>           <none>
nginx-deployment-65998d8886-qp6jj   1/1     Running   0          4m33s   10.244.1.9   host1   <none>           <none>

删除 nginx-deployment：文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl delete -f nginx.yml
deployment.extensions "nginx-deployment" deleted

 label 控制 Pod 的位置

默认配置下，Scheduler 会将 Pod 调度到所有可用的 Node。不过有些情况我们希望将 Pod 部署到指定的 Node，比如将有大量磁盘 I/O 的 Pod 部署到配置了 SSD 的 Node；或者 Pod 需要 GPU，需要运行在配置了 GPU 的节点上。文章源自靠谱运维-https://www.ixdba.net/archives/1363

Kubernetes 是通过 label 来实现这个功能的。文章源自靠谱运维-https://www.ixdba.net/archives/1363

label 是 key-value 对，各种资源都可以设置 label，灵活添加各种自定义属性。比如执行如下命令标注 k8s-node1 是配置了 SSD 的节点。文章源自靠谱运维-https://www.ixdba.net/archives/1363

第一步：定义标签文章源自靠谱运维-https://www.ixdba.net/archives/1363

disk为自定义字符串文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl label node host1 disk=ssd

第二步：查看标签文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get node --show-labels
NAME    STATUS   ROLES    AGE   VERSION   LABELS
host1   Ready    <none>   8h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,disk=ssd,kubernetes.io/hostname=host1
host2   Ready    <none>   8h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/hostname=host2
ken     Ready    master   8h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/hostname=ken,node-role.kubernetes.io/master=

disk=ssd 已经成功添加到 host1，除了 disk，Node 还有几个 Kubernetes 自己维护的 label。文章源自靠谱运维-https://www.ixdba.net/archives/1363

第三步：配置nginx.yml文章源自靠谱运维-https://www.ixdba.net/archives/1363

有了 disk 这个自定义 label，接下来就可以指定将 Pod 部署到 host1。编辑 nginx.yml：文章源自靠谱运维-https://www.ixdba.net/archives/1363

k8s运行容器之deployment（三）文章源自靠谱运维-https://www.ixdba.net/archives/1363

在 Pod 模板的 spec 里通过 nodeSelector 指定将此 Pod 部署到具有 label disktype=ssd 的 Node 上。文章源自靠谱运维-https://www.ixdba.net/archives/1363

注意：1. nodeSelector需要与containers位置保持一致文章源自靠谱运维-https://www.ixdba.net/archives/1363

2. S必须大写文章源自靠谱运维-https://www.ixdba.net/archives/1363

第四步：部署文章源自靠谱运维-https://www.ixdba.net/archives/1363

部署 Deployment文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
deployment.extensions/nginx-deployment created

第五步：查看 Pod 的运行节点文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS              RESTARTS   AGE    IP       NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-5d8db4598d-2gdmz   0/1     ContainerCreating   0          102s   <none>   host1   <none>           <none>
nginx-deployment-5d8db4598d-cq55q   0/1     ContainerCreating   0          102s   <none>   host1   <none>           <none>
nginx-deployment-5d8db4598d-qjh4x   0/1     ContainerCreating   0          102s   <none>   host1   <none>           <none>

全部 3个副本都运行在 host1 上，符合我们的预期。文章源自靠谱运维-https://www.ixdba.net/archives/1363

要删除 label disktype，执行如下命令：文章源自靠谱运维-https://www.ixdba.net/archives/1363

kubectl label node k8s-node1 disktype-文章源自靠谱运维-https://www.ixdba.net/archives/1363

– 即删除。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl label node host1 disk-
node/host1 labeled
[root@ken ~]# kubectl get node --show-labels
NAME    STATUS     ROLES    AGE   VERSION   LABELS
host1   NotReady   <none>   8h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/hostname=host1
host2   Ready      <none>   8h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/hostname=host2
ken     Ready      master   9h    v1.13.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/hostname=ken,node-role.kubernetes.io/master=

不过此时 Pod 并不会重新部署，依然在 host1 上运行。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE   IP            NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-5d8db4598d-2dbw9   1/1     Running   0          39s   10.244.1.12   host1   <none>           <none>
nginx-deployment-5d8db4598d-4brh5   1/1     Running   0          39s   10.244.1.11   host1   <none>           <none>
nginx-deployment-5d8db4598d-p87mj   1/1     Running   0          39s   10.244.1.13   host1   <none>           <none>

除非在 nginx.yml 中删除 nodeSelector 设置，然后通过 kubectl apply 重新部署。文章源自靠谱运维-https://www.ixdba.net/archives/1363

不需要删除之前的deployment,直接部署即可文章源自靠谱运维-https://www.ixdba.net/archives/1363

Kubernetes 自己会删除之前的 Pod 并调度和运行新的 Pod。文章源自靠谱运维-https://www.ixdba.net/archives/1363 

[root@ken ~]# kubectl apply -f nginx.yml
[root@ken ~]# kubectl get pod -o wide
NAME                                READY   STATUS        RESTARTS   AGE   IP            NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-5d8db4598d-p87mj   0/1     Terminating   0          2m    10.244.1.13   host1   <none>           <none>
nginx-deployment-65998d8886-t5nmv   1/1     Running       0          7s    10.244.2.9    host2   <none>           <none>
nginx-deployment-65998d8886-wz7c2   1/1     Running       0          4s    10.244.2.10   host2   <none>           <none>
nginx-deployment-65998d8886-xdlz4   1/1     Running       0          6s    10.244.1.14   host1   <none>           <none>
文章源自靠谱运维-https://www.ixdba.net/archives/1363
  • 本文由 发表于 2020年3月25日12:50:05
  • 转载请务必保留本文链接：https://www.ixdba.net/archives/1363
匿名

发表评论

匿名网友 填写信息

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen: