从Docker开发到部署到K8s全流程全解

假设我们不使用镜像仓库的情况下，在A机器上打包镜像在B机器中部署

• 打包镜像
• 导出镜像
• 部署至K8s
• 使用helm一键部署

在A机器 打包Docker镜像

• 打包镜像
• 部署镜像

在本地开发完成后，把应用程序打包成 Docker镜像 ，随后我们将一起一步步将应用部署到k8s集群。

编写Dockerfile

vi Dockerfile

FROM busybox:latest
LABEL wangligang test@email.com>
RUN echo 'hello docker'

打包镜像

docker build -t image_name -f Dockerfile

-t指定镜像名称
-f指定Dockerfile位置

查看镜像

docker images

REPOSITORY                                              TAG                 IMAGE ID            CREATED             SIZE
image_name                                   latest              1548739e1734        5s ago         20KB

导出镜像

docker save image_name > image_name.tar

#拷贝镜像到B机器
scp image_name.tar root@B:/share/

把镜像部署到B机器的minikube中

安装并启动minikube

brew install minikube
minikube start

使用NFS做为minikube的永久存储

测试环境下minikube里的内容会因每次重启而丢失，故启用NFS
在MacOs中

vi /etc/exports

写入

/share/dir -network 192.168.0.0 -mask 255.255.0.0 (rw,sync,no_root_squash)

启用nfs

sudo nfsd enable
sudo nfsd start

在minikube中使用nfs

minikube ssh
su -
mount -t nfs hostname:/share/dir /export/servers

编写deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: demo
  name: demo
spec:
  replicas: 1
  selector:
    matchLabels:
      app: demo
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        app: demo
    spec:
      containers:
      - image: image_name
        imagePullPolicy: "Never"
        name: demo
        ports:
        - containerPort: 80
        resources: {}
      restartPolicy: Always
      serviceAccountName: ""
status: {}

编写service.yaml

apiVersion: v1
kind: Service
metadata:
  labels:
    app: demo-service
  name: demo-service
spec:
  ports:
    - protocol: TCP
      port: 80
      targetPort: 80
  selector:
    app: demo
status:
  loadBalancer: {}

编写ingress.yaml

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: demo-ingress
spec:
  rules:
  - host: test.demo.com
    http:
      paths:
      - backend:
          serviceName: demo-service
          servicePort: 80

编写pv.yaml

apiVersion: v1
kind: PersistentVolume
metadata:
  name: "demo-data-pv"
  labels:
    name: demo-data-pv
    release: stable
spec:
  storageClassName: local-storage
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteMany
  persistentVolumeReclaimPolicy: Retain
  hostPath:
    path: "/export/servers"

编写pvc.yaml

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: demo-pvc
  namespace: default
spec:
  storageClassName: local-storage
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 100Mi
  selector:
    matchLabels:
      name: demo-data-pv
      release: stable
status: {}

引入Docker镜像

docker load --input /share/image_name.tar

使用kubectl部署

kubectl apply -y pv.yaml
kubectl apply -y pvc.yaml
kubectl apply -y deployment.yaml
kubectl apply -y service.yaml
kubectl apply -y ingress.yaml

查看k8s部署情况

kubectl get pods
kubectl get deployments
kubectl get services
#查看日志
kubectl logs -f --tail=10 container_namexxxxxx
#登录查看
kubectl exec -it container_namexxxxxx -- /bin/bash

使用helm一键部署

helm create demo-helm
cd demo-helm/template
rm -rf *
cp /share/yamls/deployment.yaml .
cp /share/yamls/service.yaml .
cp /share/yamls/pv.yaml .
cp /share/yamls/pvc.yaml .
cp /share/yamls/ingress.yaml .

清除上一步使用kubectl部署的资源

kubectl delete deployment --all
kubectl delete service --all
kubectl delete ingress --all
kubectl delete pvc --all
kubectl delete pv --all

一键上线

helm install demo-helm demo-helm

使用helm的一个小技巧

如果有一些项目的配置文件需要写入configmap，可以在demo-helm目录下放一个配置文件如conf.properties
在template下写一个configmap.yaml
如下

apiVersion: v1
kind: ConfigMap
metadata:
  name: demo-config
data:
  {
    {range .Files.Lines "conf.properties"}}
    {
    {.}}{
    { end }}

应用中就可以使用configmap中的变量了

一些问题

• minikube环境无法识别host里的docker image

eval $（minikube docker-env）

• deployment无法识别出同namespace里的ConfigMap中的变量值

invalid type for io.k8s.api.core.v1.ConfigMap.data: got “string”, expected “map”;

使用kubectl create configmap coding-config --from-env-file=coding-config.properties
或者如下方式,注意数字需要用引号引起来

apiVersion: v1
kind: ConfigMap
metadata:
  name: coding-config
data:
  db_host: 192.168.0.45
  db_port: "3306"
  db_name: test

• 如果 deployment指定command会覆盖Dockfile里的EXPOSEPOINT
• redis无法写入问题：MISCONF Redis is configured to save RDB snapshots

使用 sysctl vm.overcommit_memory=1修复

• minikube如果跑在docker上是不支持ingress，需要指定： minikube start --vm=true --driver=hyperkit --memory=4g

• 无法启用ingress，是因为默认minikube不支持，使用minikube addons enable ingress开启

• 远程访问mysql时开启访问别忘记指定密码grant all on *.* to 'root'@'%' identified by '123456';