基本概念

我们都知道一个pod里面可以放若干个容器，容器之间可以共享网络命名空间、文件夹。这时候一个pod里面有两个容器a和b，同时对外提供服务。如果这个两个服务之间需要彼此交互，比如a想要触发b容器的容器，或者b想要获取a容器里的一个变量之类的。我们这时候就可以使用容器共享命名空间的方法。

具体操作

先准备两个程序app1和app2

app1.go

package main

import "net/http"

func main() {
	http.HandleFunc("/", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte("this is app1\n"))
	})
	http.ListenAndServe(":8080", nil)
}

app2.go

package main

import (
	"net/http"
)

func main() {
	http.HandleFunc("/", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte("this is app2\n"))
	})
	http.ListenAndServe(":8081", nil)
}

然后分别编译丢到服务器上

编写对应的yaml:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: k8splay
  namespace: default
spec:
  replicas: 1
  selector:
    matchLabels:
      app: k8splay
  template:
    metadata:
      labels:
        app: k8splay
    spec:
      nodeName: k8s-node1
      containers:
        - name: k8splayapp1
          image: alpine:3.12
          imagePullPolicy: IfNotPresent
          command: [ "./app1" ]
          workingDir: /app
          volumeMounts:
            - name: app
              mountPath: /app
          ports:
            - containerPort: 8080
        - name: k8splayapp2
          image: alpine:3.12
          imagePullPolicy: IfNotPresent
          command: [ "./app2" ]
          workingDir: /app
          volumeMounts:
            - name: app
              mountPath: /app
          ports:
            - containerPort: 8081
      volumes:
        - name: app
          hostPath:
            path: /root/k8splay

然后直接apply

进入这个容器查看

可以看到 pid为1的进程为 ./app1，这也没啥好说的

这时候在yaml中加入 shareProcessNamespace: true，打开共享进程命名空间开关。

此时进入pod查看进程

现在任意一个容器查看 pid为1的进程就不是自己了，而且可以查看到其他容器的进程了。这就是进程命名空间共享。

现在在app1这个容器写入一个文件，在app2容器利用共享进程命名空间可以查看到了，而不需要设置emptydir

总结一下：

当启用进程命名空间共享时，容器中的进程对该pod中所有其他容器都可见
容器进程PID不再是1
容器文件系统通过 /proc/$pid/root 链接对pod中的其他容器可见

使用信号通知进程

使用命令行操作

使用脚本方式发送信号通知容器1，不通知容器2

重新修改app1.go代码：

package main

import (
	"fmt"
	"net/http"
	"os"
	"os/signal"
	"syscall"
	"time"
)

func main() {
	http.HandleFunc("/", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte("this is app1\n"))
	})
	go func() {
		for {
			ch := make(chan os.Signal)
			signal.Notify(ch, syscall.SIGTERM, syscall.SIGUSR1) // 监听 kill -15 和 kill -10 (用户自定义信号)
			c := <-ch
			fmt.Println("接收到信号:", c)
			time.Sleep(time.Millisecond * 50)
		}

	}()
	http.ListenAndServe(":8080", nil)
}

重新apply 一下

使用代码方式

app1.go

package main

import (
	"fmt"
	"net/http"
	"os"
	"os/signal"
	"syscall"
	"time"
)

var counter = 1

func main() {
	http.HandleFunc("/", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte("this is app1\n"))
	})
	http.HandleFunc("/counter", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte(fmt.Sprintf("counter: %d\n", counter)))
		counter++
	})
	go func() {
		for {
			ch := make(chan os.Signal)
			signal.Notify(ch, syscall.SIGTERM, syscall.SIGUSR1) // 监听 kill -15 和 kill -10 (用户自定义信号)
			c := <-ch
			switch c {
			case syscall.SIGUSR1:
				fmt.Println("counter被重置")
				counter = 0
			}
			time.Sleep(time.Millisecond * 50)
		}

	}()
	http.ListenAndServe(":8080", nil)
}

每次访问/counter ，counter就会加1

app2.go

package main

import (
	"net/http"
	"syscall"
)

func main() {
	http.HandleFunc("/", func(writer http.ResponseWriter, request *http.Request) {
		writer.Write([]byte("this is app2\n"))
	})
	http.HandleFunc("/rest", func(writer http.ResponseWriter, request *http.Request) {
		syscall.Kill(-1, syscall.SIGUSR1)
	})
	http.ListenAndServe(":8081", nil)
}

在app2中访问 /rest，就会发生信号给app1，从而触发counter=0的逻辑

Deploy.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: k8splay
  namespace: default
spec:
  replicas: 1
  selector:
    matchLabels:
      app: k8splay
  template:
    metadata:
      labels:
        app: k8splay
    spec:
      nodeName: k8s-node1
      shareProcessNamespace: true
      containers:
        - name: k8splayapp1
          image: byrnedo/alpine-curl
          imagePullPolicy: IfNotPresent
          command: [ "./app1" ]
          workingDir: /app
          volumeMounts:
            - name: app
              mountPath: /app
          ports:
            - containerPort: 8080
        - name: k8splayapp2
          image: byrnedo/alpine-curl
          imagePullPolicy: IfNotPresent
          command: [ "./app2" ]
          workingDir: /app
          volumeMounts:
            - name: app
              mountPath: /app
          ports:
            - containerPort: 8081
      volumes:
        - name: app
          hostPath:
            path: /root/k8splay