标签-镜像拉取策略-重启策略-pod优雅终止-pod中容器资源限制-容器类型-容器守护进程-日志排错-容器命令exec-cp

一.标签

1.概述：

标签是附加到kubernets对象（比如pod）上的键值对，标签可以在创建时附加到对象，随后也可以随时添加修改；标签不支持唯一性。

在k8s中大多数资源都是通过标签进行关联的（如pod与service）

2.标签管理

· 查看资源标签

[root@master demo]# kubectl get pods --show-labels
NAME       READY   STATUS    RESTARTS   AGE   LABELS
demo-env   1/1     Running   0          6s    <none>

· 创建标签

1.声明式创建标签

1.1编辑资源清单

[root@master demo]# cat 08-pod-nginx.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo-label
  #资源加标签
  labels:
    learn: bilibili
    k8s: wage
spec:
  containers:
  - name: demo-n
    image: nginx:1.20.1-alpine

1.2创建查看资源（标签）

[root@master demo]# kubectl apply -f 08-pod-nginx.yaml 
pod/demo-label created
[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE   LABELS
demo-label   1/1     Running   0          17s   k8s=wage,learn=bilibili

2.响应式创建标签

2.1给现有pod资源加一个标签

[root@master demo]# kubectl label pods demo-label time=1255
pod/demo-label labeled
[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE     LABELS
demo-label   1/1     Running   0          4m11s   k8s=wage,learn=bilibili,time=1255

注：响应式创建的标签，如果重新拉取pod（删除后再创建pod），标签会消失

3.删除标签

[root@master demo]# kubectl label pods demo-label time-
pod/demo-label unlabeled
[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE    LABELS
demo-label   1/1     Running   0          8m2s   k8s=wage,learn=bilibili
[root@master demo]#

注：声明式创建的标签（资源清单），删除后，重新拉区，标签会重新创建

4.修改标签

[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE    LABELS
demo-label   1/1     Running   0          8m2s   k8s=wage,learn=bilibili
[root@master demo]# kubectl label pods demo-label k8s=kubernets --overwrite
pod/demo-label labeled
[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE   LABELS
demo-label   1/1     Running   0          11m   k8s=kubernets,learn=bilibili

5.通过标签删除pod

[root@master demo]# kubectl get pods --show-labels 
NAME         READY   STATUS    RESTARTS   AGE   LABELS
demo-label   1/1     Running   0          11m   k8s=kubernets,learn=bilibili
[root@master demo]# kubectl delete pods -l k8s
pod "demo-label" deleted
[root@master demo]# kubectl get pods -o wide
No resources found in default namespace.

二.pod镜像拉取策略

· Always：对比所有镜像来源，选择时间最新的

· Never：只使用本地镜像，本地若没有，也不会去远程仓库拉取

· IfNotPresent（默认策略）：先看本地，本地有选择本地，本地没有，选择远程仓库

1.编写镜像拉取策略的资源清单

[root@master demo]# cat 09-pod-pull.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo-pull
  labels:
    demo: pull
spec:
  containers:
  - name: nginx
    image: nginx:1.20.1-alpine
    #设置镜像拉取策略
    imagePullPolicy: IfNotPresent

[root@master demo]# kubectl apply -f 09-pod-pull.yaml 
pod/demo-pull created
[root@master demo]# kubectl get pods -o wide
NAME        READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo-pull   1/1     Running   0          12s   10.100.1.30   worker1   <none>           <none>

三.pod中容器的重启策略

· Always（默认值）：当容器退出时（不论什么原因），自动拉起新的容器；

· Never：当容器退出时（不论什么原因），都不会重新创建拉起新的容器；

· OnFailure：当容器“意外退出”时，才会拉起新的容器；

1.容器重启

· 编写容器重启策略的资源清单

[root@master demo]# cat 10-pod-restart.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo-restart
  labels:
    demo: restart
spec:
  #设置容器重启策略
  restartPolicy: Always
  containers:
  - name: nginx
    image: nginx:1.20.1-alpine

· 创建查看资源

[root@master demo]# kubectl apply -f 10-pod-restart.yaml 
pod/demo-restart created
[root@master demo]# kubectl get pods -o wide
NAME           READY   STATUS    RESTARTS   AGE     IP            NODE      NOMINATED NODE   READINESS GATES
demo-restart   1/1     Running   0          10s     10.100.2.34   worker2   <none>           <none>

· 在worker节点删除容器测试是否会重新拉起容器

[root@worker2 ~]# docker ps | grep "nginx"
714d17c65cef   7f18bdc92ca5                                        "/docker-entrypoint.…"   About a minute ago   Up About a minute             k8s_nginx_demo-restart_default_03b7f2a9-6a09-4259-aa71-b9e5a186dc08_0
[root@worker2 ~]# docker rm -f 714d17c65cef
714d17c65cef
[root@worker2 ~]# docker ps | grep "nginx"
d4b6f0852149   7f18bdc92ca5                                        "/docker-entrypoint.…"   2 seconds ago        Up 2 seconds                  k8s_nginx_demo-restart_default_03b7f2a9-6a09-4259-aa71-b9e5a186dc08_1

· master节点查看pod状态

此时，pod的restarts从0变为1

[root@master demo]# kubectl get pods -o wide
NAME           READY   STATUS    RESTARTS   AGE    IP            NODE      NOMINATED NODE   READINESS GATES
demo-restart   1/1     Running   1          113s   10.100.2.34   worker2   <none>           <none>

四.pod的优雅终止

在pod的删除时，系统会自动延迟30s，为了给pod处理未处理完的请求

1.编辑资源清单

[root@master demo]# cat 11-pod-stop.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo-stop
  labels:
    demo: stop
spec:
  #pod优雅终止字段，定义延时kill信号的时间，给pod处理未完成的请求时间；
  #缓期多少秒时间执行；若不设置，默认是30s；
  terminationGracePeriodSeconds: 5
  containers:
  - name: demo-nginx
    image: nginx:1.20.1-alpine
    #定义容器的生命周期（容器启动做什么动作，容器停止前做什么动作）
    lifecycle:
      #容器启动前做什么
      postStart:
        exec:
          command:
          - "sh"
          - "-c"
          - "echo \"postStart at $(date +%F_%T)\" >> /poststart.log"
      #容器停止前做什么
      preStop:
        exec:
          command:
          - "sh"
          - "-c"
          - "echo \"prestop at $(date +%F_%T)\" >> /prestop.log"

2.创建查看资源

[root@master demo]# kubectl apply -f 11-pod-stop.yaml 
pod/demo-stop created
[root@master demo]# kubectl get pods -o wide
NAME        READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo-stop   1/1     Running   0          9s    10.100.1.31   worker1   <none>           <none>

3.进入容器查看是否完成容器启动前置动作

[root@master demo]# kubectl exec demo-stop -it -- sh
/ # ls
bin                   etc                   mnt                   root                  sys
dev                   home                  opt                   run                   tmp
docker-entrypoint.d   lib                   poststart.log         sbin                  usr
docker-entrypoint.sh  media                 proc                  srv                   var
/ # cat poststart.log 
postStart at 2024-07-14_05:32:59
/ # exit

五.pod中容器的资源限制

给pod设置固定的硬件使用限制，例如cpu、磁盘、内存

CPU资源单位：Kubernetes将1CPU以1000m来表示，CPU的最小资源单位为m，1m表示千分之一CPU。通常一个容器使用的CPU配额为100m~ 300m。

1.编辑资源清单

[root@master demo]# cat 12-pod-resources.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo-res
  labels:
    demo: res
spec:
  containers:
  - name: demo
    image: nginx:1.20.1-alpine
    #设置资源限制
    resources:
      #最大资源的使用限制；最多就能用这么多的资源；
      limits:
        #2c;
        cpu: 2000m
        #限制内存
        memory: 40M
      #期望资源限制；需要宿主机预留的资源，我可以不用，但是你必须要有
      requests:
        #1c
        cpu: 1000m
        #限制内存
        memory: 20M

2.创建查看资源

[root@master demo]# kubectl apply -f 12-pod-resources.yaml 
pod/demo-res created
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo-res   1/1     Running   0          8s    10.100.2.35   worker2   <none>           <none>

3.查看容器的资源大小是否为设置阈值

[root@worker2 ~]# docker ps | grep demo
0a1df97e8497   7f18bdc92ca5                                        "/docker-entrypoint.…"   35 seconds ago   Up 35 seconds             k8s_demodemo-res_default_a8af141d-f1cd-4474-95d1-eb3405413cfe_0
4c204b53ce98   registry.aliyuncs.com/google_containers/pause:3.6   "/pause"                 36 seconds ago   Up 35 seconds             k8s_POD_demo-res_default_a8af141d-f1cd-4474-95d1-eb3405413cfe_0
[root@worker2 ~]# docker stats 0a1df97e8497

六.容器类型

· 基础架构容器【pause】：运行pod中的容器时，提供容器的网络名称空间

· 初始化容器【initContainers】：

完成一些业务容器运行前的操作，如执行命令，如果初始化容器没有创建成功，将一直重启，业务容器也就无法创建出来；

它可以延后业务容器的启动时间；

1.基础架构容器pause

查看基础架构容器

[root@worker2 ~]# docker ps | grep pause

2.初始化容器initContainers

· 编辑带有初始化容器的pod资源清单

[root@master demo]# cat 13-pod-init.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: m12
spec:
  #声明初始化容器
  initContainers:
  - name: init-demo1
    image: alpine
    #在初始化容器中执行命令
    command:
    - "sleep"
    - "5"
  - name: init-demo2
    image: alpine
    #在初始化容器中执行命令 
    command:
    - "sleep"
    - "5"
  #业务容器
  containers:
  - name: demoinit
    image: alpine
    #给容器一个标准输入，也就是守护进程
    stdin: true

·创建查看资源

[root@master demo]# kubectl apply -f 13-pod-init.yaml 
pod/m12 created
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS     RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
m12    0/1     Init:0/2   0          12s   <none>   worker1   <none>           <none>

·查看状态

[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS     RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m12    0/1     Init:0/2   0          24s   10.100.1.32   worker1   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS     RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m12    0/1     Init:1/2   0          27s   10.100.1.32   worker1   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS            RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m12    0/1     PodInitializing   0          52s   10.100.1.32   worker1   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m12    1/1     Running   0          63s   10.100.1.32   worker1   <none>           <none>

七.pod中容器的守护进程

1.创建资源清单，拉取alpine最小linux系统

[root@master demo]# cat 14-pod-dae.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: m-alpine
spec:
  containers:
    - name: c1
      image: alpine

2.创建查看资源

查看pod资源，发现，启动后会结束，因为没有守护进程

[root@master demo]# kubectl apply -f 14-pod-dae.yaml 
pod/m-alpine created
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS              RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
m-alpine   0/1     ContainerCreating   0          7s    <none>   worker1   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS      RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m-alpine   0/1     Completed   0          19s   10.100.1.33   worker1   <none>           <none>

3.修改资源清单，加入标准输入stdin

[root@master demo]# cat 14-pod-dae.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: m-alpine
spec:
  containers:
    - name: c1
      image: alpine
      stdin: true

4，再次查看pod

[root@master demo]# kubectl apply -f 14-pod-dae.yaml 
pod/m-alpine created
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS              RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
m-alpine   0/1     ContainerCreating   0          7s    <none>   worker2   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS              RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
m-alpine   0/1     ContainerCreating   0          17s   <none>   worker2   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME       READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
m-alpine   1/1     Running   0          18s   10.100.2.36   worker2   <none>           <none>

5.命令方式守护进程

·command

1.编辑资源清单

[root@master demo]# cat 14-pod-dae.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo
spec:
  containers:
    - name: c1
      image: alpine
      command:
      - "tail"
      - "-f"
      - "/etc/hosts"

2.创建查看资源

[root@master demo]# kubectl apply -f 14-pod-dae.yaml 
pod/demo created
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS              RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
demo   0/1     ContainerCreating   0          6s    <none>   worker2   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo   1/1     Running   0          19s   10.100.2.37   worker2   <none>           <none>

·args

1.编辑资源清单

[root@master demo]# cat 14-pod-dae.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo
spec:
  containers:
    - name: c1
      image: alpine
      args:
      - "tail"
      - "-f"
      - "/etc/hosts"

2.创建查看资源

[root@master demo]# kubectl apply -f 14-pod-dae.yaml 
pod/demo created
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS              RESTARTS   AGE   IP       NODE      NOMINATED NODE   READINESS GATES
demo   0/1     ContainerCreating   0          5s    <none>   worker2   <none>           <none>
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo   1/1     Running   0          83s   10.100.2.38   worker2   <none>           <none>

· command与args结合(Dockerfile的命令讲解)

args可以当做command的参数进行命令执行；

跟docker中的守护进程命令一样；command就类似于ENTRYPOINT；args就类似于CMD；

[root@master pod]# cat 14-pod-dae.yaml 

apiVersion: v1
kind: Pod
metadata:
  name: demo
spec:
  containers:
    - name: c1
      image: alpine
      command:
      - "tail"
      - "-f"
      args:
      - "/etc/hosts"

[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo   1/1     Running   0          83s   10.100.2.38   worker2   <none>           <none>

八.pod排障之二-日志查询

1.查看pod

[root@master demo]# cat 14-pod-dae.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: demo
spec:
  containers:
    - name: demo
      image: nginx:1.20.1-alpine

[root@master demo]# kubectl apply -f 14-pod-dae.yaml 
pod/demo created
[root@master demo]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
demo   1/1     Running   0          6s    10.100.2.39   worker2   <none>           <none>

2.模拟循环访问pod

while true ; 
do 
  curl 10.100.2.39; 
  sleep 0.5 
done

3.查看pod日志

[root@master demo]# kubectl logs -f demo

4.面试题

如果pod中容器发生重启，如何查看重启前的容器的log日志信息？

· 模拟容器重启

[root@worker2 ~]# docker ps | grep demo
[root@worker2 ~]# docker kill ad9a8501d12c 
ad9a8501d12c

· 验证是否重新拉起容器

· 查看容器重启前，上一个容器的日志信息

-p

[root@master demo]# kubectl logs -p demo

九.进入pod容器exec与cp命令

#进入pod中容器

kubectl  exec  pod名称  -it  --  sh

#拷贝容器中数据到宿主机（默认pod中第一个容器）

kubectl  cp  pod名称:/root/demo.txt    ./

#拷贝宿主机数据到容器

kubectl  cp  ./a.log   pod名称:/mnt/

#指定容器拷贝

kubectl  cp -c 容器名  pod名称:/root/demo.txt    ./

#pod外部执行命令

[root@master pod]# kubectl exec demo -it -- ifconfig