Docker 开发最佳实践
Docker development best practices
The following development patterns have proven to be helpful for people building applications with Docker. If you have discovered something we should add, let us know.
How to keep your images small
Small images are faster to pull over the network and faster to load into memory when starting containers or services. There are a few rules of thumb to keep image size small:
Start with an appropriate base image. For instance, if you need a JDK, consider basing your image on the official
openjdk
image, rather than starting with a genericubuntu
image and installingopenjdk
as part of the Dockerfile.Use multistage builds. For instance, you can use the
maven
image to build your Java application, then reset to thetomcat
image and copy the Java artifacts into the correct location to deploy your app, all in the same Dockerfile. This means that your final image doesn’t include all of the libraries and dependencies pulled in by the build, but only the artifacts and the environment needed to run them.If you need to use a version of Docker that does not include multistage builds, try to reduce the number of layers in your image by minimizing the number of separate
RUN
commands in your Dockerfile. You can do this by consolidating multiple commands into a singleRUN
line and using your shell’s mechanisms to combine them together. Consider the following two fragments. The first creates two layers in the image, while the second only creates one.RUN apt-get -y update
RUN apt-get install -y python
RUN apt-get -y update && apt-get install -y python
If you have multiple images with a lot in common, consider creating your own base image with the shared components, and basing your unique images on that. Docker only needs to load the common layers once, and they are cached. This means that your derivative images use memory on the Docker host more efficiently and load more quickly.
To keep your production image lean but allow for debugging, consider using the production image as the base image for the debug image. Additional testing or debugging tooling can be added on top of the production image.
When building images, always tag them with useful tags which codify version information, intended destination (
prod
ortest
, for instance), stability, or other information that is useful when deploying the application in different environments. Do not rely on the automatically-createdlatest
tag.
Where and how to persist application data
- Avoid storing application data in your container’s writable layer using storage drivers. This increases the size of your container and is less efficient from an I/O perspective than using volumes or bind mounts.
- Instead, store data using volumes.
- One case where it is appropriate to use bind mounts is during development, when you may want to mount your source directory or a binary you just built into your container. For production, use a volume instead, mounting it into the same location as you mounted a bind mount during development.
- For production, use secrets to store sensitive application data used by services, and use configs for non-sensitive data such as configuration files. If you currently use standalone containers, consider migrating to use single-replica services, so that you can take advantage of these service-only features.
Use swarm services when possible
- When possible, design your application with the ability to scale using swarm services.
- Even if you only need to run a single instance of your application, swarm services provide several advantages over standalone containers. A service’s configuration is declarative, and Docker is always working to keep the desired and actual state in sync.
- Networks and volumes can be connected and disconnected from swarm services, and Docker handles redeploying the individual service containers in a non-disruptive way. Standalone containers need to be manually stopped, removed, and recreated to accommodate configuration changes.
- Several features, such as the ability to store secrets and configs, are only available to services rather than standalone containers. These features allow you to keep your images as generic as possible and to avoid storing sensitive data within the Docker images or containers themselves.
- Let
docker stack deploy
handle any image pulls for you, instead of usingdocker pull
. This way, your deployment doesn’t try to pull from nodes that are down. Also, when new nodes are added to the swarm, images are pulled automatically.
There are limitations around sharing data amongst nodes of a swarm service. If you use Docker for AWS or Docker for Azure, you can use the Cloudstor plugin to share data amongst your swarm service nodes. You can also write your application data into a separate database which supports simultaneous updates.
Use CI/CD for testing and deployment
When you check a change into source control or create a pull request, use Docker Cloud or another CI/CD pipeline to automatically build and tag a Docker image and test it. Docker Cloud can also deploy tested apps straight into production.
Take this even further with Docker EE by requiring your development, testing, and security teams to sign images before they can be deployed into production. This way, you can be sure that before an image is deployed into production, it has been tested and signed off by, for instance, development, quality, and security teams.
Differences in development and production environments
Development | Production |
---|---|
Use bind mounts to give your container access to your source code. | Use volumes to store container data. |
Use Docker for Mac or Docker for Windows. | Use Docker EE if possible, with userns mapping for greater isolation of Docker processes from host processes. |
Don’t worry about time drift. | Always run an NTP client on the Docker host and within each container process and sync them all to the same NTP server. If you use swarm services, also ensure that each Docker node syncs its clocks to the same time source as the containers. |
Docker 开发最佳实践的更多相关文章
- web前端开发最佳实践笔记
一.文章开篇 由于最近也比较忙,一方面是忙着公司的事情,另外一方面也是忙着看书和学习,所以没有时间来和大家一起分享知识,现在好了,终于回归博客园的大家庭了,今天我打算来分享一下关于<web前端开 ...
- [转]Android开发最佳实践
——欢迎转载,请注明出处 http://blog.csdn.net/asce1885 ,未经本人同意请勿用于商业用途,谢谢—— 原文链接:https://github.com/futurice/and ...
- Hadoop MapReduce开发最佳实践(上篇)
body{ font-family: "Microsoft YaHei UI","Microsoft YaHei",SimSun,"Segoe UI& ...
- 【读书笔记】iOS-微信公众平台开发最佳实践
一,微信是由腾讯公司广州研发中心产品团队开发,该团队经理张小龙被称为“微信之父”,公司总裁马化腾确定该产品名称为“微信”. 二,常见问题及解决方案. 1,请求URL超时. 这种情况一般是由于服务器网速 ...
- Android开发最佳实践《IT蓝豹》
Android开发最佳实践 移动开发Android经验分享应用GoogleMaterial Design 摘要:前 段时间,Google公布了Android开发最佳实践的一系列课程,涉及到一些平时 ...
- Android和PHP开发最佳实践
Android和PHP开发最佳实践 <Android和PHP开发最佳实践>基本信息作者: 黄隽实丛书名: 移动应用开发技术丛书出版社:机械工业出版社ISBN:9787111410508上架 ...
- iOS应用开发最佳实践
<iOS应用开发最佳实践> 基本信息 作者: 王浩 出版社:电子工业出版社 ISBN:9787121207679 上架时间:2013-7-22 出版日期:2013 年8月 开本:16 ...
- 【社区公益】送《Web前端开发最佳实践》给需要的人
算起来至今,我进入软件开发行业已经有11年之久.从最初的研究人工智能,到后来的Web开发,控件开发,直到现在纯粹的Web前端开发.虽然没有大的作品问世,但也是勤勤恳恳,踏实做事,低调做人.从来不吹牛逼 ...
- Web前端开发最佳实践系列文章汇总
Web前端开发最佳实践(1):前端开发概述 Web前端开发最佳实践(2):前端代码重构 Web前端开发最佳实践(3):前端代码和资源的压缩与合并 Web前端开发最佳实践(4):在页面中添加必要的met ...
随机推荐
- Extjs4前端开发代码规范参考
准则: 一致性, 隔离与统一管理, 螺旋式重构改进, 消除重复, 借鉴现有方案 1. 保证系统实现的一致性,寻求一致性方案, 相同或相似功能尽量用统一模式处理: 2. 尽可能使用隔离技术 ...
- 20165305 实验二:Java面向对象程序设计
2-1 参考 http://www.cnblogs.com/rocedu/p/6371315.html#SECUNITTEST 参考http://www.cnblogs.com/rocedu/p/67 ...
- 2、在VM上的 CentOS 6.5 上安装mysql
1.查看系统是否安装了MySQL 使用命令: #rpm -qa | grep mysql 2.卸载已安装的MySQL 卸载mysql命令如下: #rpm -e - ...
- 大数据自学1-CentOS 下安装CDH及Cloudera Manager
前面花了一段时间将Ubuntu,Hadoop装完,装到Hbase时,发现Hbase 与Hadoop是有兼容性问题的,Hbase 2.1版是不支持Hadoop 3.11版的,怪不得装起来那么多问题了. ...
- 为什么List.add()所增加的数据都是一样的
1. 先上代码: List<Person> list = new ArrayList<>(); Person p = new Person(); try { Class.for ...
- django的url配置
在一个请求到达的时候,最先达到的就是视图层,然后根据url映射到视图函数.这一部分我们来说明url的配置. 概述 为了给一个应用设计URL,你需要创建一个Python 模块,通常称为URLconf(U ...
- script 跳出小窗口
sss
- JavaScript数组实现图片轮播
最终效果 注:图片来源于百度图片 文件结构: 代码: <!DOCTYPE html> <html> <head> <meta charset="UT ...
- P5290 [十二省联考2019]春节十二响(堆+启发式合并)
P5290 [十二省联考2019]春节十二响 从特殊到一般 我们先看链的情况. 我们把点$1$左右的两条子链分别扔入堆里 每次取出两个堆的最大值,把答案累加上更大的那个(另一堆为空则直接加上去). 那 ...
- 定制FileField中的上传文件名称
FileField中的upload_to属性可以设定上传文件的存储目录和名称,它可以是个字符串,也可以是个callable,比如一个方法. 当upload_to的值设为一个方法时,就可以对上传文件的名 ...