In daily work we may have used many kinds of docker images. For me I pretty like the jupyter images which are well maintained.

 $ docker run -p 8888:8888 jupyter/scipy-notebook
 

Here "jupyter/scipy-notebook" is the image name. How can I customize and release an image for my own app? I will show you the guide step by step.

First, we create a project dir whose name is "docker-proj". There are two important stuff in it.

• Dockerfile
• App files

App file is anything you just want to run in docker. For instance, mine in this case is a perl script. This script uses a module from CPAN, Gmail::Mailbox::Validate, which checks if the given username has a valid mailbox on gmail.

The demo code for this app are,

 use Gmail::Mailbox::Validate;
 use strict;
     
 my $username = shift || die "$0 username\n";
 my $v = Gmail::Mailbox::Validate->new();
 print $username,
       $v->validate($username) ? " exists\n" : " not exists\n";
 

Save the code in script "gmbox.pl", and run it you will see the output as follows.

 $ perl gmbox.pl 
 gmbox.pl username
    
 $ perl gmbox.pl lisaleung
 lisaleung exists
 

Put the script into dir "docker-proj". And, create Dockerfile in the same dir with the following content.

 FROM perl:5.34
 RUN cpanm Gmail::Mailbox::Validate && mkdir -p /usr/src/app
 COPY gmbox.pl /usr/src/app
 WORKDIR /usr/src/app
 ENTRYPOINT [ "perl", "./gmbox.pl" ]
 

Some comments on above configuration,

• FROM - we are using image perl 5.34 as base image
• RUN - run the commands to install modules from CPAN and create a work dir
• COPY - copy app files from local to docker image
• WORKDIR - step into work dir
• ENTRYPOINT - the command should be run when docker instance starts up

In current dir where the subdir "docker-proj" is located, run the following command to build an image.

 $ docker build -t gmbox docker-proj/
 

Here "gmbox" is the image name, which you can check by this command.

 $ docker image ls
 REPOSITORY               TAG       IMAGE ID       CREATED         SIZE
 gmbox                    latest    38fca2257430   2 minutes ago   894MB
 

Now please go to dockerhub.com creating an account. You will use this account ID to build the official image.

For instance, my account name is "geekml", then the official name for my image is "geekml/gmbox".

Let's tag the image we just made in local,

 $ docker tag gmbox geekml/gmbox:latest
 

Log into dockerhub in terminal,

 $ docker login
 

Input the account name and password you have created in dockerhub.com to login.

The last step is to push the tagged image to dockerhub.

 $ docker push geekml/gmbox:latest
 The push refers to repository [docker.io/geekml/gmbox]
 b4c22e3253a2: Pushed 
 ...
 latest: digest: sha256:9fdead1956eeb93735f8301cbd34440f9383969d9729336d0d29f567c027a41a size: 2212
 

So you can run this docker image from everywhere. For instance, in another host just pull the image,

 $ docker pull geekml/gmbox
 Using default tag: latest
 latest: Pulling from geekml/gmbox
 ...
 18f60cd0e987: Pull complete 
 Digest: sha256:9fdead1956eeb93735f8301cbd34440f9383969d9729336d0d29f567c027a41a
 Status: Downloaded newer image for geekml/gmbox:latest
 docker.io/geekml/gmbox:latest
 

And run the image as a container,

 $ docker run geekml/gmbox johnlee
 johnlee exists
 

Check the image you have pulled down,

 $ docker image list
 REPOSITORY     TAG       IMAGE ID       CREATED          SIZE
 geekml/gmbox   latest    38fca2257430   17 minutes ago   894MB
 

All jobs done well now.