Dockerfile directives

Dockerfile directives#

Here is a closer look on different docker directives with examples.

Execute with run#

There are two dockerfile directives that allow to execute something with start of docker container: CMD and ENTRYPOINT.

Here is example of the image that modifies alpine in way to print message Hello world during container start.

cat << EOF > dockerfile
FROM alpine
CMD ["echo", "'Hello world'"]
EOF

docker build -t temp_image . &> /dev/null
docker run --rm temp_image

docker rmi temp_image &> /dev/null
rm dockerfile

'Hello world'

Parametrisation#

With the ENV and ARG directives, you can define values to be reused during different phases of the build. There are some differences between the two; refer to the Docker documentation page.

Environment variables#

For creating environment variable you can use ENV dockerfile instruction. Feature of this diretive is that defined variable will be included to the environment of the resulting containers.

In the following example, a Dockerfile is created using ENV to define TEST_VAR with the value "env_variable", and a container is started based on the resulting image.

cat << EOF > /tmp/dockerfile
FROM alpine
ENV TEST_VAR="env_variable"
WORKDIR \$TEST_VAR
EOF

docker build -t test_image -f /tmp/dockerfile . &> /dev/null
docker run -itd --name env_example --rm test_image &> /dev/null

Variable used in dockerfile as name for new directory. The following cell checks if the variable was actually created.

docker exec env_example pwd

/env_variable

But you can still use variables from the container. The following cell demonstrates this by showing the env of the container in question.

docker exec env_example env

PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
HOSTNAME=a2b1510d2ee2
TEST_VAR=env_variable
HOME=/root

The following cell clears the environment after experiments.

docker stop env_example &> /dev/null
docker rmi test_image &> /dev/null

Build arguments#

Build arguments that can be created with ARG directive in nginx, it is a variable that can be used for build command for dockerfile but there is no information about it after build.

The following cell builds image and runs container based on it, by properties of gotten container we’ll learn properties of the variable defined with ARG directive.

cat << EOF > /tmp/dockerfile
FROM alpine
ARG TEST_VAR="arg_variable"
WORKDIR \$TEST_VAR
EOF

docker build -t test_image -f /tmp/dockerfile . &> /dev/null
docker run -itd --name arg_example --rm test_image &> /dev/null

The variable is used as a directory name for the WORKDIR directive, as a result the initial folder of the container is the same as the value of the variable.

docker exec arg_example pwd

/arg_variable

But there are no such variables in the environment of the resulting container, which is shown by the following cell.

docker exec arg_example env

PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
HOSTNAME=2123ad7acb59
HOME=/root

Don’t forget to clean up the environment.

docker stop arg_example &> /dev/null
docker rmi test_image &> /dev/null

Specify in build#

You can specify the argument value during the build with the --build-arg <arg name>=<value> argument of to the docker build.

The following example builds a dockerfile which working directory depends on the TEST_VAR argument, and builds the corresponding image with --build-args TEST_VAR=name_for_folder.

cat << EOF > /tmp/dockerfile
FROM alpine
ARG TEST_VAR
WORKDIR \$TEST_VAR
EOF

docker build --build-arg TEST_VAR=name_for_folder -t test_image -f /tmp/dockerfile . &> /dev/null

The following cell shows the WORKDIR of the resulting image corresponding to the value of the TEST_VAR.

docker run --rm test_image pwd

/name_for_folder

Cleaning the environment.

docker rmi test_image &> /dev/null