无法触发自定义（手动）Bitbucket Pipeline，yml报错缺少import属性

问题场景 #

• 自动Pipeline的执行步骤可正常运行，但手动触发自定义staging-deploy流水线时失败
• 流水线预期执行流程：
  1. 构建镜像并推送到DockerHub
  2. 通过SSH连接到DO Droplet，拉取镜像并运行容器

错误提示 #

There is an error in your bitbucket-pipelines.yml at [pipelines > custom > staging-deploy]. Required attributes are not set: import

原配置代码 #

image: atlassian/default-image:3

pipelines:
  custom:
    staging-deploy:
      deploy:
        - step:
          name: Deploy to DockerHub
          services:
            - docker
          script:
            - export IMAGE_NAME=$DOCKER_HUB_USERNAME/$PROJECT_NAME:$BITBUCKET_COMMIT
            - docker build -t $IMAGE_NAME .
            - docker login --username $DOCKER_HUB_USERNAME --password $DOCKER_HUB_PASSWORD
            - docker push $IMAGE_NAME
        
        - step:
          name: Deploy to Droplet
          script:
            - export IMAGE_NAME=$DOCKER_HUB_USERNAME/$PROJECT_NAME:$BITBUCKET_COMMIT
            - pipe: atlassian/ssh-run:0.6.1
              variables:
                SSH_USER: $SSH_USER
                SERVER: $SSH_SERVER
                COMMAND: 'docker login --username $DOCKER_HUB_USERNAME --password $DOCKER_HUB_PASSWORD; docker pull $IMAGE_NAME; docker stop $PROJECT_NAME; docker rm $PROJECT_NAME; docker run -d --restart always --name $PROJECT_NAME -p $SERVER_PORT:$SERVER_PORT $IMAGE_NAME; docker image prune -a -f'

问题原因及修复方案 #

自定义Pipeline的语法规则和自动Pipeline不同，不能在自定义流水线节点下直接使用deploy关键字。自动Pipeline中的deploy是用于指定部署阶段的专属关键字，但自定义流水线要求直接将步骤列表挂载在流水线名称节点下，不需要额外的层级包裹。

修改后的正确配置：

image: atlassian/default-image:3

pipelines:
  custom:
    staging-deploy:
      - step:
          name: Deploy to DockerHub
          services:
            - docker
          script:
            - export IMAGE_NAME=$DOCKER_HUB_USERNAME/$PROJECT_NAME:$BITBUCKET_COMMIT
            - docker build -t $IMAGE_NAME .
            - docker login --username $DOCKER_HUB_USERNAME --password $DOCKER_HUB_PASSWORD
            - docker push $IMAGE_NAME
        
      - step:
          name: Deploy to Droplet
          script:
            - export IMAGE_NAME=$DOCKER_HUB_USERNAME/$PROJECT_NAME:$BITBUCKET_COMMIT
            - pipe: atlassian/ssh-run:0.6.1
              variables:
                SSH_USER: $SSH_USER
                SERVER: $SSH_SERVER
                COMMAND: 'docker login --username $DOCKER_HUB_USERNAME --password $DOCKER_HUB_PASSWORD; docker pull $IMAGE_NAME; docker stop $PROJECT_NAME; docker rm $PROJECT_NAME; docker run -d --restart always --name $PROJECT_NAME -p $SERVER_PORT:$SERVER_PORT $IMAGE_NAME; docker image prune -a -f'

核心修改点：移除staging-deploy节点下的deploy层级，直接将两个step作为列表项挂载在staging-deploy下，符合自定义流水线的结构要求。

内容的提问来源于stack exchange，提问作者Alooza