近期在进行Vue项目自动化构建，遇到一些问题，下面记录一个典型的问题。
1. 问题出现 在使用docker部署vue项目的时候，使用了multi-stage-build来构建最终镜像，然鹅在构建时却出现了如下问题：
1 2 3 4 5 6 7 8 Building nginx_manager Step 1/12 : FROM node:lts-alpine as base Service 'nginx_manager' failed to build: Error parsing reference: "node:lts-alpine as base" \ is not a valid repository/tag: invalid reference format SSH: EXEC: completed after 1,636 ms SSH: Disconnecting configuration [tencent_server] ... ERROR: Exception when publishing, exception message [Exec exit status not zero. Status [1]] Finished: UNSTABLE
这贴一下我最原始的Dockerfile和docker-compose.yml
1 2 3 4 5 6 7 8 9 10 11 12 13 FROM node:lts-alpine as base-build WORKDIR /app COPY package*.json ./ RUN npm install COPY . . RUN npm run build:prod # production stage FROM nginx:stable-alpine as nginx-build # 将修改的文件替换掉容器中的nginx配置 # COPY default.conf /etc/nginx/conf.d/default.conf COPY --from=base-build /app/dist /usr/share/nginx/html EXPOSE 80 CMD ["nginx", "-g", "daemon off;"]
1 2 3 4 5 6 7 8 9 10 11 version: "3" services: nginx: container_name: manager build: context: ./manager dockerfile: Dockerfile ports: - 8080:80 environment: HOST: 0.0.0.0
2. 查找解决方案 于是开展各种搜索，搜索到一个相关问题，推荐使用target的方式来指定具体的构建阶段
相关链接：https://stackoverflow.com/questions/53093487/multi-stage-build-in-docker-compose
修改后的docker-compose.yml
1 2 3 4 5 6 7 8 9 10 11 12 version: "3.4" services: nginx_manager: container_name: manager build: context: ./manager dockerfile: Dockerfile target: nginx-build ports: - 8080:80 environment: HOST: 0.0.0.0
执行后依旧是上面的错误。。。
于是又进行了搜索，在GitHub上搜索到相关问题，原因竟然在于docker版本过低！
相关链接：https://github.com/tootsuite/mastodon/issues/8240
GitHub上的讨论指出：
1 Docker's "multi stage build" with keyword "as" is supported on Docker 17.05 or later.
查看服务器docker版本，确实是低于17.05，于是重装了docker。
1 2 3 4 5 6 7 8 9 # 安装依赖 yum install -y yum-utils device-mapper-persistent-data lvm2 # 修改为阿里云的源 yum-config-manager --add-repo http://mirrors.aliyun.com/docker-ce/linux/centos/docker-ce.repo # 删除原来的docker yum remove docker docker-client docker-client-latest docker-common docker-latest docker-latest-logrotate \ docker-logrotate docker-selinux docker-engine-selinux docker-engine # 安装docker ce yum install docker-ce
安装成功后启动docker，结果出现如下错误：
1 2 Error response from daemon: Unknown runtime specified docker-runc Error: failed to start containers: 7253e1be7fb3
真是一波未平一波又起啊，不过有解决方案：
1 2 3 4 5 6 7 8 9 10 11 12 13 [root@VM_12_255_centos ~]# cd /var/lib/docker/containers/ [root@VM_12_255_centos containers]# grep -rl "docker-runc" . ./550dc5fda621f02b5079e0d70d502c01752fef6a03f0bf3dbf533f0634fc6013/hostconfig.json ./d2bc485ef863b5742c29d4e95e5de641aaf5b474657eb16c47b51bc3d679ca1a/hostconfig.json ./afa79db1bd41284e99446404afee1f43f06a7381889fd90961493d83ffb31658/hostconfig.json ./b405ade18866912841727c384c4eb71625d59865dc72676c39dfdde1592c8748/hostconfig.json ./9c291209f3e09e5cdf6f7333873d8216be531aba940d185f12160a1ff98fdc6e/hostconfig.json ./9c291209f3e09e5cdf6f7333873d8216be531aba940d185f12160a1ff98fdc6e/config.v2.json ./e5a7f498b6a8501f4e625e4e8598b819577141dbb137f945356777d3c9c0c077/hostconfig.json ./c58152c4ceeb6506aeca1eacf029c51e53cfc3848bc588120b51e97ee6463f74/hostconfig.json ./b4b59541feb03b0f102f521da336f358be4c70f23ecdc837644e3835c024535d/hostconfig.json ./b4b59541feb03b0f102f521da336f358be4c70f23ecdc837644e3835c024535d/config.v2.json [root@VM_12_255_centos containers]# sed -i "s/docker-runc//g" ./550dc5fda621f02b5079e0d70d502c01752fef6a03f0bf3dbf533f0634fc6013/hostconfig.json
将上面grep搜索出来的文件全都执行sed命令替换，之后正常重启即可
相关链接：https://forums.docker.com/t/after-upgrade-to-docker-ce-containers-will-not-start/40243/3
3. 总结 至此，此次问题告一段落，版本兼容问题确实不容小觑啊！
Tips 欢迎收藏和转发，感谢你的支持！(๑•̀ㅂ•́)و✧
欢迎关注我：后端小哥，专注后端开发，希望和你一起进步！
