Image¶

Every statement like RUN, COPY or ADD in the dockerfile will add a layer into docker image.

run image¶

docker run -it <image-id>                        #from default entrypoint
docker run -it --entrypoint /bin/bash <image-id> #from another entrypoint

list images with sort¶

docker images | sort -k1 -h #sort by col1 - repo

remove image with multiple repos/tags¶

docker rmi -f <image-id>
docker rmi [repo-name1]:[tag1] [repo-name2]:[tag2]

check docker image layer sizes¶

docker history <image-id>
docker history --no-trunc <image-id> #keep full created-by command

reduce docker image size¶

https://jcristharif.com/conda-docker-tips.html

https://uwekorn.com/2021/03/01/deploying-conda-environments-in-docker-how-to-do-it-right.html

• Use an official base image whenever possible

• Use variations of Alpine images when possible

• Every RUN instruction in the Dockerfile writes a new layer in the image

• Every layer requires extra space on disk

• To keep the number layers to a minimum, commands should be combined into a single RUN instruction

• Free spaces after update using apt clean all

• Use multistage builds

• Use .dockerignore to exclude files not needed

• Squash layers: https://github.com/goldmann/docker-squash

Conda docker image size¶

https://pythonspeed.com/articles/conda-docker-image-size/

Use conda-pack to remove conda and use multistage build together to reduce image size.

The base Conda environment is necessary for installation of packages, but once we’re running the code it really doesn’t add much.

multiple versions of package in the same image, why???¶