Dockerfile

A Dockerfile is a plain-text file describing the steps to build an image, typically for your own application. To build a Docker image from a Dockerfile in the current directory, enter:

docker image build -t <image_name> .

The period at the end of the command references the current directory. Append -f <file> to use an alternative path or file name.

The commands below will be used most often.

# comment

Lines beginning with # denote a comment:

#my comment

ARG arguments

Variables can be passed at build time with the --build-arg <name>=<value> option. The value is imported with an ARG statement:

#get myvar
ARG myvar

A default value can be defined when nothing is passed:

#get myvar, with default of "myimage"
ARG myvar=myimage

Use its value elsewhere by prepending a $ symbol:

RUN echo $myvar

Where the value must be set in part of a string, a ${name} substitution can be defined:

FROM ${myvar}:latest

ENV environment variables

Set an environment variable:

ENV HOME=/home/app

Use its value elsewhere by prepending $ symbol or ${} container:

RUN echo $HOME
RUN echo ${HOME}

FROM <image> starting image

Create a new image using an existing image as a starting point. This will usually be the first command in a Dockerfile:

#latest Node LTS on Alpine Linux
FROM node:lts-alpine

WORKDIR working directory

Set the working directory for any following COPY, ADD, RUN, CMD and ENTRYPOINT instructions:

WORKDIR /home/myapp

COPY files from the host to an image

Using file patterns:

#copy all files to current folder
COPY . .

#copy all txt files to /doc/ folder
COPY *.txt /doc/

#copy all files and assign ownership to a user and group
#(Linux containers only)
COPY --chown=myuser:mygroup . .

ADD files

ADD is similar to COPY, but it also supports using URLs and tar files as the source. This may be useful, although using a RUN with chained curl and tar commands will create a smaller Docker image with fewer layers.

Mount a VOLUME

By creating a mount point we can make files and folders accessible to other containers, e.g., in the following the data in the folder /data/myvol is available to other containers in the network too,

RUN mkdir -p /data/myvol
VOLUME /data/myvol

Multiple volumes or mount points can be specified as:

VOLUME /myvol1 /myvol2 /myvol3

Example use: If you are serving a Node.js application to NGINX and you want to share some Node.js files with NGINX, you can mount the client-side HTML, CSS, JavaScript, and media files in a static folder on a Node.js container. Now these files can be served directly by an NGINX web server running in another container.

Set a USER

Define the user (and optionally a group) to use for RUN, CMD and ENTRYPOINT instructions:

USER myuser

RUN a command

RUN issues instructions during the build, such as installation or configuration commands. Any number is permitted in the Dockerfile.

Execute commands using shell form:

RUN npm install

Or an exec (array) form to run an executable directly, which is the recommended option:

RUN ["npm", "install"]

📌 The Exec form correctly receives Unix signals such as the signal to terminate. It is the most commonly used and is also recommended over the Shell form. The Shell omits weird array notations. Docker executes the command via a command shell. It prefixes the given command with /bin/sh -c, so in our case, it runs /bin/sh -c node ./index.js. This form is avoided because it does not pass signals to its subprocesses, which causes issues later in the termination of the server.

EXPOSE a port

Informs Docker that the container will listen on a specified network port. Note that the exposed port must be published with -p when using docker run:

EXPOSE 3000

CMD execute container

CMD sets a default application start instruction if one isn’t specified when launching the container with docker run. Only one is permitted in the Dockerfile, so only the last is processed.

Commands can be executed using shell form:

CMD node ./index.js

Or an exec (array) form to directly run the executable:

CMD ["node", "/index.js"]

In general, use CMD to provide a default command that can be overridden on the command line when the container is launched.

ENTRYPOINT execute container

ENTRYPOINT sets a start instruction when building an executable Docker image. The application will run when the container is launched. Only one is permitted in the Dockerfile and it overrides any CMD instructions.

Commands can be executed using shell form:

ENTRYPOINT node ./index.js

or exec (array) form – the recommended option – to directly run the executable:

ENTRYPOINT ["node", "/index.js"]

.dockerignore file patterns

.docekerignore specifies filename patterns to omit when using COPY and ADD. For example:

Get hands-on with 1200+ tech skills courses.