如何修复K8s Pod内执行docker compose up时的Docker daemon连接问题

我希望在Kubernetes Pod内执行docker compose up命令启动Docker Compose，以下是我的Deployment配置：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: test-system
  labels:
    app: test
    name: test
spec:
  replicas: 1
  selector:
    matchLabels:
      app: test
  template:
    metadata:
      labels:
        app: test
    spec:  
      # init-container 
      initContainers:
      - name: init-docker-compose
        image: k8s-testsystem-init:origin-main-stable
        imagePullPolicy: Always
        securityContext:
          runAsUser: 0
          runAsGroup: 0
        command: ["/bin/sh", "-c"]
        args: ["cp -R /opt/docker /data && chmod -R 777 /data/docker && cat /tmp/.env >> /data/docker/.env"]
        volumeMounts:
          - name: compose-files
            mountPath: /data
          - name: docker-config
            mountPath: /tmp
      containers:
      # docker 
      - name: docker-test
        image: docker:latest
        command: ["/bin/sh", "-c"]
        args: ["cd /data/docker && docker commpose up"]
        securityContext:
           privileged: true
        ports:
        - containerPort: 80
        - containerPort: 443
        - containerPort: 5005
        - containerPort: 502
        volumeMounts:
          - name: compose-files
            mountPath: /data
      # volumes
      volumes:
        # modbus profiles volumes
        - name: compose-files
          emptyDir: {}
        - name: docker-config
          configMap: 
            name: docker-configmap
            items: 
              - key: env
                path: .env

创建该资源后Pod退出并报错：

Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

但如果移除该命令，通过shell手动启动则一切正常，请问该如何修复这个问题？

问题核心是：直接执行docker compose up时，容器内的Docker daemon（dockerd）还未启动；手动进shell操作正常，是因为你大概率先手动启动了daemon进程，再执行compose命令。

具体修改步骤 #

1. 修正命令拼写错误：配置里的docker commpose up多了一个m，改为docker compose up。
2. 先启动Docker daemon并等待就绪：调整启动命令，先让dockerd后台运行，再等待它完全就绪后执行compose命令。

修改后的容器配置片段：

containers:
- name: docker-test
  image: docker:latest
  command: ["/bin/sh", "-c"]
  # 用循环检查替代固定sleep，确保daemon就绪后再启动compose
  args: ["dockerd & while ! docker info > /dev/null 2>&1; do sleep 1; done && cd /data/docker && docker compose up"]
  securityContext:
     privileged: true
  ports:
  - containerPort: 80
  - containerPort: 443
  - containerPort: 5005
  - containerPort: 502
  volumeMounts:
    - name: compose-files
      mountPath: /data

说明 #

• dockerd &：让Docker daemon在后台运行，不阻塞后续命令执行。
• while ! docker info > /dev/null 2>&1; do sleep 1; done：循环执行docker info检查，直到daemon能正常响应，比固定sleep 10更可靠，避免因环境差异导致的启动不充分问题。

内容的提问来源于stack exchange，提问作者Christian