Managing files in your image

Introduction to Docker

Tim Sangster

Software Engineer @ DataCamp

COPYing files into an image

The COPY instruction copies files from our local machine into the image we're building:

COPY <src-path-on-host> <dest-path-on-image>
COPY /projects/pipeline_v3/pipeline.py /app/pipeline.py

docker build -t pipeline:v3 .
...
[4/4] COPY ./projects/pipeline_v3/pipeline.py /app/pipeline.py

If the destination path does not have a filename, the original filename is used:

COPY /projects/pipeline_v3/pipeline.py /app/

COPYing folders

Not specifying a filename in the src-path will copy all the file contents.

COPY <src-folder> <dest-folder>
COPY /projects/pipeline_v3/ /app/

COPY /projects/pipeline_v3/ /app/ will copy everything under pipeline_v3/:

/projects/
    pipeline_v3/
        pipeline.py
        requirements.txt
        tests/
            test_pipeline.py

Copy files from a parent directory

/init.py
/projects/
    Dockerfile
    pipeline_v3/
        pipeline.py

If our current working directory is in the projects/ folder.

We can't copy init.py into an image.

docker build -t pipeline:v3 .
 => ERROR [4/4] COPY ../init.py /     0.0s
failed to compute cache key: "../init.py" not found: not found

Downloading files

Instead of copying files from a local directory, files are often downloaded in the image build:

Download a file

RUN curl <file-url> -o <destination>

Unzip the file

RUN unzip <dest-folder>/<filename>.zip

Remove the original zip file

RUN rm <copy_directory>/<filename>.zip

Downloading files efficiently

Each instruction that downloads files adds to the total size of the image.
Even if the files are later deleted.
The solution is to download, unpack and remove files in a single instruction.

RUN curl <file_download_url> -o <destination_directory>/<filename>.zip \
&& unzip <destination_directory>/<filename>.zip -d <unzipped-directory> \
&& rm <destination_directory>/<filename>.zip

Summary

Usage	Dockerfile Instruction
Copy files from host to the image	COPY <src-path-on-host> <dest-path-on-image>
Copy a folder from host to the image	COPY <src-folder> <dest-folder>
We can't copy from a parent directory where we build a Dockerfile	~~COPY ../<file-in-parent-directory> /~~

Keep images small by downloading, unzipping, and cleaning up in a single RUN instruction:

RUN curl <file_download_url> -o <destination_directory> \
&& unzip <destination_directory>/<filename>.zip -d <unzipped-directory> \
&& rm <destination_directory>/<filename>.zip

Let's practice!

Introduction to Docker