Have the static tagging mechanism of docker bugged you and prevented you from creating flawless and incrementally versioned CD/CI of docker images? Well then you may like this tool.
You keep a docker tag template and a version in a configuration file and you use this toolkit, which can automatically increment the version stored in the configuration file when building and tagging the image, and automatically pick the full tag when saving and deploying the image. How does it help? Well you combine it with git and you got yourself a flawless CD/CI without any hardcoded tags.
You'd need docker(duh), python3 and pip3 (for python3). Simply clone the repository and do a
pip install dove-docker
inside the cloned repository.
Like docker, it's a command-line tool. Do a dove on the shell to display all the help.
Usage: dove [OPTIONS] COMMAND [ARGS]...
A docker versioning extension that manages docker tags through a JSON
file, so that the user doesn't have to get into the hassle of writing and
updating image tags.
Maintained by: Intech A&I
For more info, visit:
https://github.com/intech-iiot/dove
Options:
--help Show this message and exit.
Commands:
build Build the docker image with saved tag
bump Just bump up the current version
get Get the current tag from the config
new Initialize a new dove config
push Push the image with tag saved in the config
reset Reset the version at position(s) to 0
save Save an image with the tag from the config
tag Tag another image with the saved tag
The tool requires a dove.json configuration file to be placed with the Dockerfile. This file can be created by executing the following on shell:
dove new -t my-repo/my-image:{0}.{1}.{2} -i 3.2.1
The -t puts a tag template with string positional parameters and the -i puts an initial version for the image.
After creating the configuration file, run the following in the directory with rhe configuration file:
dove get
This will show my-repo/my-image:3.2.1 for the above configuration. Evidently, the version is split using a dot(.) and the components are substituted on the positional parameters. Ofcourse you can use the tag in other commands such as:
docker image inspect $(dove get)
For this, a dockerfile must be present with the dove configuration. Do a:
dove build
This will start building the docker image with the current tag from the configuration file. To bump up the version, do a:
dove build --bump 0
This will increment the portion of the version that is supposed to replace the positional parameter at {0}. Similarly, to bump up multiple portions, do a:
dove build --bump 0 --bump 1
if you're inclined to use the docker cli commands yourself, you can still use the configuration and its version bumping features by bumping the version manually. To do so, do a:
dove bump --position 0
This will update the version in the configuration and return the bumped up tag in the stdout, which can be nested in a docker command itself e.g.
docker build -t $(dove bump --position 0) ./
or you can simply get the tag afterwards
docker build -t $(dove get) ./
You can also reset the version number at any position by:
dove reset --position 0
This will reset the version number at position 0, which is useful if you follow a major-minor release cycle since a major release bump means a reset of the minor release version to 0.
An existing image can be tagged with a dove configuration. To do so, do a:
dove tag -s source-image-tag
The source-image will be tagged with the tag inferred from the configuration file in the current context
To save the image with the inferred tag from the configuration, do a:
dove save -f filename
If the tag format contained a remote repo, I also have implemented a push function. To do so, simply do a:
dove push
The dove build and dove push commands can be provided with docker build and docker push command-line arguments through the --args or -a argument. For example:
dove build -a --pull
or
dove build -a --pull -a --no-cache -a --label=pickle-rick
Yeah I know, it's pretty repetitive, and you can only use the long version with =, but I didn't know any other way of separating the dove and docker command line args. At least it's better than nothing.
From version 0.0.3 onwards commands and parameters for maintaining alternate tag formats was introduced. What this means is that the dove.json can contain more than one tag format in which the version is substituted. This is done by keeping a primary tag format and maintaining a name-value pair of alternate tag formats. These tags can be used in the other commands by means of a -l or --alt flag. This is very useful for multi-stage builds. You can have a different tag with the same version for different builds by using this.
To add a new alternate tag to the existing dove.json configuration, do a:
dove add-alt -l my-alt-tag my-repo/my-image:{0}.{1}.{2}
To view the alternates that were added to the configuration file, do a:
dove alts
An alternate can be removed through it's name by doing a:
dove remove-alt -n my-alt-tag
Any command that uses tags from the configuration file has an existing flag -l or --alt which is used to specify the name of the alternate you itend to use. For example, to build an image with an alternate tag, do a:
dove build -l my-alt-tag
I'm strictly not a python programmer and really did it as a side project so I can make some pretty streamlined CD/CI. I'd really appreciate if you want to contribute to this little utility of mine.