When Docker[1] exploded onto the scene a few years ago, it brought containers and container images to the masses. Although Linux containers existed before then, Docker made it easy to get started with a user-friendly command-line interface and an easy-to-understand way to build images using the Dockerfile format. But while it may be easy to jump in, there are still some nuances and tricks to building container images that are usable, even powerful, but still small in size.
First pass: Clean up after yourself
Some of these examples involve the same kind of cleanup you would use with a traditional server, but more rigorously followed. Smaller image sizes are critical for quickly moving images around, and storing multiple copies of unnecessary data on disk is a waste of resources. Consequently, these techniques should be used more regularly than on a server with lots of dedicated storage.
An example of this kind of cleanup is removing cached files from an image to recover space. Consider the difference in size between a base image with Nginx[2] installed by dnf with and without the metadata and yum cache cleaned up:
FROM fedora:28
LABEL maintainer Chris Collins
RUN dnf install -y nginx
-----
# Dockerfile w/o cache
FROM fedora:28
LABEL maintainer Chris Collins <
RUN dnf install -y nginx \
&& dnf clean all \
&& rm -rf /var/cache/yum
-----
[chris@krang] $ docker build -t cache -f Dockerfile .
[chris@krang] $ docker images --format "{{.Repository}}: {{.Size}}"
| head -n 1
cache: 464 MB
[chris@krang] $ docker build -t no-cache -f Dockerfile-wo-cache .
[chris@krang] $ docker images --format "{{.Repository}}: {{.Size}}" | head -n 1