Best Practices for Writing Dockerfiles

Best practices for writing Dockerfiles

Overview

Docker can build images automatically by reading the instructions from aDockerfile, a text file that contains all the commands, in order, needed tobuild a given image. Dockerfiles adhere to a specific format and use aspecific set of instructions. You can learn the basics on theDockerfile Reference page. Ifyou’re new to writing Dockerfiles, you should start there.

This document covers the best practices and methods recommended by Docker,Inc. and the Docker community for creating easy-to-use, effectiveDockerfiles. We strongly suggest you follow these recommendations (in fact,if you’re creating an Official Image, you must adhere to these practices).

You can see many of these practices and recommendations in action in the buildpack-deps Dockerfile.

Note: for more detailed explanations of any of the Dockerfile commandsmentioned here, visit the Dockerfile Reference page.

General guidelines and recommendations

Containers should be ephemeral

The container produced by the image your Dockerfile defines should be asephemeral as possible. By “ephemeral,” we mean that it can be stopped anddestroyed and a new one built and put in place with an absolute minimum ofset-up and configuration.

Use a .dockerignore file

In most cases, it’s best to put each Dockerfile in an empty directory. Then,add to that directory only the files needed for building the Dockerfile. Toincrease the build’s performance, you can exclude files and directories byadding a .dockerignore file to that directory as well. This file supportsexclusion patterns similar to .gitignore files. For information on creating one,see the .dockerignore file.

Avoid installing unnecessary packages

In order to reduce complexity, dependencies, file sizes, and build times, youshould avoid installing extra or unnecessary packages just because theymight be “nice to have.” For example, you don’t need to include a text editorin a database image.

Run only one process per container

In almost all cases, you should only run a single process in a singlecontainer. Decoupling applications into multiple containers makes it mucheasier to scale horizontally and reuse containers. If that service depends onanother service, make use of container linking.

Minimize the number of layers

You need to find the balance between readability (and thus long-termmaintainability) of the Dockerfile and minimizing the number of layers ituses. Be strategic and cautious about the number of layers you use.

Sort multi-line arguments

Whenever possible, ease later changes by sorting multi-line argumentsalphanumerically. This will help you avoid duplication of packages and make thelist much easier to update. This also makes PRs a lot easier to read andreview. Adding a space before a backslash (\) helps as well.

Here’s an example from the buildpack-deps image:

RUN apt-get update && apt-get install -y \
  bzr \
  cvs \
  git \
  mercurial \
  subversion

Build cache

During the process of building an image Docker will step through theinstructions in your Dockerfile executing each in the order specified.As each instruction is examined Docker will look for an existing image in itscache that it can reuse, rather than creating a new (duplicate) image.If you do not want to use the cache at all you can use the --no-cache=trueoption on the docker build command.

However, if you do let Docker use its cache then it is very important tounderstand when it will, and will not, find a matching image. The basic rulesthat Docker will follow are outlined below:

• Starting with a base image that is already in the cache, the nextinstruction is compared against all child images derived from that baseimage to see if one of them was built using the exact same instruction. Ifnot, the cache is invalidated.

• In most cases simply comparing the instruction in the Dockerfile with oneof the child images is sufficient. However, certain instructions requirea little more examination and explanation.

• For the ADD and COPY instructions, the contents of the file(s)in the image are examined and a checksum is calculated for each file.The last-modified and last-accessed times of the file(s) are not considered inthese checksums. During the cache lookup, the checksum is compared against thechecksum in the existing images. If anything has changed in the file(s), suchas the contents and metadata, then the cache is invalidated.

• Aside from the ADD and COPY commands, cache checking will not look at thefiles in the container to determine a cache match. For example, when processinga RUN apt-get -y update command the files updated in the containerwill not be examined to determine if a cache hit exists. In that case justthe command string itself will be used to find a match.

Once the cache is invalidated, all subsequent Dockerfile commands willgenerate new images and the cache will not be used.

The Dockerfile instructions

Below you’ll find recommendations for the best way to write thevarious instructions available for use in a Dockerfile.

FROM

Dockerfile reference for the FROM instruction

Whenever possible, use current Official Repositories as the basis for yourimage. We recommend the Debian imagesince it’s very tightly controlled and kept extremely minimal (currently under100 mb), while still being a full distribution.

RUN

Dockerfile reference for the RUN instruction

As always, to make your Dockerfile more readable, understandable, andmaintainable, split long or complex RUN statements on multiple lines separatedwith backslashes.

apt-get

Probably the most common use-case for RUN is an application of apt-get. TheRUN apt-get command, because it installs packages, has several gotchas to lookout for.

You should avoid RUN apt-get upgrade or dist-upgrade, as many of the“essential” packages from the base images won’t upgrade inside an unprivilegedcontainer. If a package contained in the base image is out-of-date, you shouldcontact its maintainers.If you know there’s a particular package, foo, that needs to be updated, useapt-get install -y foo to update automatically.

Always combine RUN apt-get update with apt-get install in the same RUNstatement, for example:

    RUN apt-get update && apt-get install -y \
        package-bar \
        package-baz \
        package-foo

Using apt-get update alone in a RUN statement causes caching issues andsubsequent apt-get install instructions fail.For example, say you have a Dockerfile:

    FROM ubuntu:14.04
    RUN apt-get update
    RUN apt-get install -y curl

After building the image, all layers are in the Docker cache. Suppose you latermodify apt-get install by adding extra package:

    FROM ubuntu:14.04
    RUN apt-get update
    RUN apt-get install -y curl nginx

Docker sees the initial and modified instructions as identical and reuses thecache from previous steps. As a result the apt-get update is NOT executedbecause the build uses the cached version. Because the apt-get update is notrun, your build can potentially get an outdated version of the curl and nginxpackages.

Using RUN apt-get update && apt-get install -y ensures your Dockerfileinstalls the latest package versions with no further coding or manualintervention. This technique is known as “cache busting”. You can also achievecache-busting by specifying a package version. This is known as version pinning,for example:

    RUN apt-get update && apt-get install -y \
        package-bar \
        package-baz \
        package-foo=1.3.*

Version pinning forces the build to retrieve a particular version regardless ofwhat’s in the cache. This technique can also reduce failures due to unanticipated changesin required packages.

Below is a well-formed RUN instruction that demonstrates all the apt-getrecommendations.

RUN apt-get update && apt-get install -y \
    aufs-tools \
    automake \
    build-essential \
    curl \
    dpkg-sig \
    libcap-dev \
    libsqlite3-dev \
    lxc=1.0* \
    mercurial \
    reprepro \
    ruby1.9.1 \
    ruby1.9.1-dev \
    s3cmd=1.1.* \
 && apt-get clean \
 && rm -rf /var/lib/apt/lists/*

The s3cmd instructions specifies a version 1.1.0*. If the image previouslyused an older version, specifying the new one causes a cache bust of apt-getupdate and ensure the installation of the new version. Listing packages oneach line can also prevent mistakes in package duplication.

In addition, cleaning up the apt cache and removing /var/lib/apt/lists helpskeep the image size down. Since the RUN statement starts withapt-get update, the package cache will always be refreshed prior toapt-get install.

CMD

Dockerfile reference for the CMD instruction

The CMD instruction should be used to run the software contained by yourimage, along with any arguments. CMD should almost always be used in theform of CMD [“executable”, “param1”, “param2”…]. Thus, if the image is for aservice (Apache, Rails, etc.), you would run something likeCMD ["apache2","-DFOREGROUND"]. Indeed, this form of the instruction isrecommended for any service-based image.

In most other cases, CMD should be given an interactive shell (bash, python,perl, etc), for example, CMD ["perl", "-de0"], CMD ["python"], orCMD [“php”, “-a”]. Using this form means that when you execute something likedocker run -it python, you’ll get dropped into a usable shell, ready to go.CMD should rarely be used in the manner of CMD [“param”, “param”] inconjunction with ENTRYPOINT, unlessyou and your expected users are already quite familiar with how ENTRYPOINTworks.

EXPOSE

Dockerfile reference for the EXPOSE instruction

The EXPOSE instruction indicates the ports on which a container will listenfor connections. Consequently, you should use the common, traditional port foryour application. For example, an image containing the Apache web server woulduse EXPOSE 80, while an image containing MongoDB would use EXPOSE 27017 andso on.

For external access, your users can execute docker run with a flag indicatinghow to map the specified port to the port of their choice.For container linking, Docker provides environment variables for the path fromthe recipient container back to the source (ie, MYSQL_PORT_3306_TCP).

ENV

Dockerfile reference for the ENV instruction

In order to make new software easier to run, you can use ENV to update thePATH environment variable for the software your container installs. Forexample, ENV PATH /usr/local/nginx/bin:$PATH will ensure that CMD [“nginx”]just works.

The ENV instruction is also useful for providing required environmentvariables specific to services you wish to containerize, such as Postgres’sPGDATA.

Lastly, ENV can also be used to set commonly used version numbers so thatversion bumps are easier to maintain, as seen in the following example:

ENV PG_MAJOR 9.3
ENV PG_VERSION 9.3.4
RUN curl -SL http://example.com/postgres-$PG_VERSION.tar.xz | tar -xJC /usr/src/postgress && …
ENV PATH /usr/local/postgres-$PG_MAJOR/bin:$PATH

Similar to having constant variables in a program (as opposed to hard-codingvalues), this approach lets you change a single ENV instruction toauto-magically bump the version of the software in your container.

ADD or COPY

Dockerfile reference for the ADD instruction
Dockerfile reference for the COPY instruction

Although ADD and COPY are functionally similar, generally speaking, COPYis preferred. That’s because it’s more transparent than ADD. COPY onlysupports the basic copying of local files into the container, while ADD hassome features (like local-only tar extraction and remote URL support) that arenot immediately obvious. Consequently, the best use for ADD is local tar fileauto-extraction into the image, as in ADD rootfs.tar.xz /.

If you have multiple Dockerfile steps that use different files from yourcontext, COPY them individually, rather than all at once. This will ensure thateach step’s build cache is only invalidated (forcing the step to be re-run) if thespecifically required files change.

For example:

COPY requirements.txt /tmp/
RUN pip install /tmp/requirements.txt
COPY . /tmp/

Results in fewer cache invalidations for the RUN step, than if you put theCOPY . /tmp/ before it.

Because image size matters, using ADD to fetch packages from remote URLs isstrongly discouraged; you should use curl or wget instead. That way you candelete the files you no longer need after they’ve been extracted and you won’thave to add another layer in your image. For example, you should avoid doingthings like:

ADD http://example.com/big.tar.xz /usr/src/things/
RUN tar -xJf /usr/src/things/big.tar.xz -C /usr/src/things
RUN make -C /usr/src/things all

And instead, do something like:

RUN mkdir -p /usr/src/things \
    && curl -SL http://example.com/big.tar.xz \
    | tar -xJC /usr/src/things \
    && make -C /usr/src/things all

For other items (files, directories) that do not require ADD’s tarauto-extraction capability, you should always use COPY.

ENTRYPOINT

Dockerfile reference for the ENTRYPOINT instruction

The best use for ENTRYPOINT is to set the image’s main command, allowing thatimage to be run as though it was that command (and then use CMD as thedefault flags).

Let’s start with an example of an image for the command line tool s3cmd:

ENTRYPOINT ["s3cmd"]
CMD ["--help"]

Now the image can be run like this to show the command’s help:

$ docker run s3cmd

Or using the right parameters to execute a command:

$ docker run s3cmd ls s3://mybucket

This is useful because the image name can double as a reference to the binary asshown in the command above.

The ENTRYPOINT instruction can also be used in combination with a helperscript, allowing it to function in a similar way to the command above, evenwhen starting the tool may require more than one step.

For example, the Postgres Official Imageuses the following script as its ENTRYPOINT:

#!/bin/bash
set -e

if [ "$1" = 'postgres' ]; then
    chown -R postgres "$PGDATA"

    if [ -z "$(ls -A "$PGDATA")" ]; then
        gosu postgres initdb
    fi

    exec gosu postgres "$@"
fi

exec "$@"

Note:This script uses the exec Bash commandso that the final running application becomes the container’s PID 1. This allowsthe application to receive any Unix signals sent to the container.See the ENTRYPOINThelp for more details.

The helper script is copied into the container and run via ENTRYPOINT oncontainer start:

COPY ./docker-entrypoint.sh /
ENTRYPOINT ["/docker-entrypoint.sh"]

This script allows the user to interact with Postgres in several ways.

It can simply start Postgres:

$ docker run postgres

Or, it can be used to run Postgres and pass parameters to the server:

$ docker run postgres postgres --help

Lastly, it could also be used to start a totally different tool, such as Bash:

$ docker run --rm -it postgres bash

VOLUME

Dockerfile reference for the VOLUME instruction

The VOLUME instruction should be used to expose any database storage area,configuration storage, or files/folders created by your docker container. Youare strongly encouraged to use VOLUME for any mutable and/or user-serviceableparts of your image.

USER

Dockerfile reference for the USER instruction

If a service can run without privileges, use USER to change to a non-rootuser. Start by creating the user and group in the Dockerfile with somethinglike RUN groupadd -r postgres && useradd -r -g postgres postgres.

Note: Users and groups in an image get a non-deterministicUID/GID in that the “next” UID/GID gets assigned regardless of imagerebuilds. So, if it’s critical, you should assign an explicit UID/GID.

You should avoid installing or using sudo since it has unpredictable TTY andsignal-forwarding behavior that can cause more problems than it solves. Ifyou absolutely need functionality similar to sudo (e.g., initializing thedaemon as root but running it as non-root), you may be able to use“gosu”.

Lastly, to reduce layers and complexity, avoid switching USER backand forth frequently.

WORKDIR

Dockerfile reference for the WORKDIR instruction

For clarity and reliability, you should always use absolute paths for yourWORKDIR. Also, you should use WORKDIR instead of proliferatinginstructions like RUN cd … && do-something, which are hard to read,troubleshoot, and maintain.

ONBUILD

Dockerfile reference for the ONBUILD instruction

An ONBUILD command executes after the current Dockerfile build completes.ONBUILD executes in any child image derived FROM the current image. Thinkof the ONBUILD command as an instruction the parent Dockerfile givesto the child Dockerfile.

A Docker build executes ONBUILD commands before any command in a childDockerfile.

ONBUILD is useful for images that are going to be built FROM a givenimage. For example, you would use ONBUILD for a language stack image thatbuilds arbitrary user software written in that language within theDockerfile, as you can see in Ruby’s ONBUILD variants.

Images built from ONBUILD should get a separate tag, for example:ruby:1.9-onbuild or ruby:2.0-onbuild.

Be careful when putting ADD or COPY in ONBUILD. The “onbuild” image willfail catastrophically if the new build’s context is missing the resource beingadded. Adding a separate tag, as recommended above, will help mitigate this byallowing the Dockerfile author to make a choice.

Examples for Official Repositories

These Official Repositories have exemplary Dockerfiles:

• Go
• Perl
• Hy
• Rails

Additional resources:

• Dockerfile Reference
• More about Base Images
• More about Automated Builds
• Guidelines for Creating OfficialRepositories