如何在Kubernetes中向卷挂载的ConfigMap脚本传递环境变量

下面提供几种可行的实现方式，根据你的场景需求选择即可：

方法1：容器启动时直接替换变量 #

直接在Workflow的容器命令中加入sed替换逻辑，读取传入的参数替换ConfigMap文件中的$VAR：

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: test-wf-
spec:
  entrypoint: main
  # 定义可外部传入的参数
  arguments:
    parameters:
      - name: var-value
        value: "hello-argo" # 默认值，可动态传入
  templates:
    - name: main
      inputs:
        parameters:
          - name: var-value
      container:
        image: "ubuntu"
        command: ["/bin/sh", "-c"]
        args:
          - |
            # 替换文件中的$VAR为传入的参数值
            sed -i "s/\$VAR/{{inputs.parameters.var-value}}/" /mnt/vc/test
            # 验证替换结果
            cat /mnt/vc/test
        volumeMounts:
          - name: vc
            mountPath: "/mnt/vc"
      volumes:
        - name: vc
          configMap:
            name: test-pod-cfg
            items:
              - key: test-pod.yaml
                path: test

这种方式简单直接，适合临时替换、不需要保留替换后ConfigMap的场景。

方法2：用Init容器提前完成替换 #

如果需要在主容器启动前就完成变量替换，可借助Init容器处理原始ConfigMap文件，再将处理后的文件挂载给主容器：

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: test-wf-
spec:
  entrypoint: main
  arguments:
    parameters:
      - name: var-value
        value: "init-demo"
  templates:
    - name: main
      inputs:
        parameters:
          - name: var-value
      volumes:
        - name: processed-cfg
          emptyDir: {} # 存储替换后的文件
        - name: original-cfg
          configMap:
            name: test-pod-cfg
            items:
              - key: test-pod.yaml
                path: test
      # 先执行Init容器完成变量替换
      initContainers:
        - name: replace-var
          image: "ubuntu"
          command: ["/bin/sh", "-c"]
          args:
            - |
              sed "s/\$VAR/{{inputs.parameters.var-value}}/" /mnt/original/test > /mnt/processed/test
          volumeMounts:
            - name: original-cfg
              mountPath: "/mnt/original"
            - name: processed-cfg
              mountPath: "/mnt/processed"
      container:
        image: "ubuntu"
        command: ["/bin/sh", "-c", "cat /mnt/vc/test"]
        volumeMounts:
          - name: processed-cfg
            mountPath: "/mnt/vc"

Init容器会先处理ConfigMap内容，主容器挂载的是已经完成替换的文件，适合主容器逻辑无需关心替换步骤的场景。

方法3：动态生成替换后的ConfigMap #

如果需要生成持久化的、已替换变量的ConfigMap，可以通过脚本动态创建新的ConfigMap，再挂载给主容器：

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  name: test-wf-dynamic-cfg
spec:
  entrypoint: main
  arguments:
    parameters:
      - name: var-value
        value: "dynamic-cfg-demo"
  templates:
    - name: main
      inputs:
        parameters:
          - name: var-value
      steps:
        - - name: generate-cfgmap
            template: create-dynamic-cfg
        - - name: use-cfgmap
            template: mount-dynamic-cfg
            arguments:
              parameters:
                - name: cfg-name
                  value: "{{steps.generate-cfgmap.outputs.parameters.cfg-name}}"
    # 动态生成带替换后内容的ConfigMap
    - name: create-dynamic-cfg
      script:
        image: "bitnami/kubectl:latest"
        command: [sh]
        source: |
          # 生成替换VAR后的Pod配置
          cat <<EOF > test-pod.yaml
          apiVersion: v1
          kind: Pod
          metadata:
            name: test-pod
          spec:
            containers:
              - name: test
                image: ubuntu
                command: ["/busybox/sh", "-c", "echo {{inputs.parameters.var-value}}"]
          EOF
          # 创建新的ConfigMap
          kubectl create configmap dynamic-test-cfg --from-file=test-pod.yaml
          # 输出ConfigMap名称供后续步骤使用
          echo -n "dynamic-test-cfg"
      outputs:
        parameters:
          - name: cfg-name
            valueFrom:
              path: /tmp/script-output
    # 挂载动态生成的ConfigMap
    - name: mount-dynamic-cfg
      inputs:
        parameters:
          - name: cfg-name
      container:
        image: "ubuntu"
        command: ["/bin/sh", "-c", "cat /mnt/vc/test"]
        volumeMounts:
          - name: vc
            mountPath: "/mnt/vc"
      volumes:
        - name: vc
          configMap:
            name: "{{inputs.parameters.cfg-name}}"
            items:
              - key: test-pod.yaml
                path: test

这种方式适合需要复用替换后配置的场景，生成的ConfigMap可被其他资源引用。

内容的提问来源于stack exchange，提问作者Karamchand