How To Contribute

Basics

  1. Our source code is hosted on OpenDev Kolla Git. Bugs should be filed on launchpad.

  2. Please follow OpenStack Gerrit Workflow to contribute to Kolla.

  3. Note the branch you’re proposing changes to. master is the current focus of development. Kolla project has a strict policy of only allowing backports in stable/branch, unless when not applicable. A bug in a stable/branch will first have to be fixed in master.

  4. Please file a blueprint of kolla for any significant code change and a bug for any significant bug fix. See how to reference a bug or a blueprint in the commit message. For simple changes, contributors may optionally add the text “TrivialFix” to the commit message footer to indicate to reviewers a bug is not required.

  5. We use a whiteboard to keep track of CI gate status, release status, stable backports, planning and feature development status.

Please use the existing sandbox repository, available at sandbox, for learning, understanding and testing the Gerrit Workflow.

Adding a release note

All new features should have a documented release note. To add a release note run the following command:

tox -e venv -- reno new <feature-being-added>

Typically in this project we do not add release notes for bug fixes. Upgrade notes can be extremely helpful for operators so these are encouraged.

Adding a new service

Kolla aims to both containerise and deploy all services within the OpenStack ecosystem. This is a constantly moving target as the ecosystem grows, so these guidelines aim to help make adding a new service to Kolla a smooth experience.

The image

Kolla follows Best practices for writing Dockerfiles when designing and implementing services where at all possible.

We use jinja2 templating syntax to help manage the volume and complexity that comes with maintaining multiple Dockerfiles for multiple different base operating systems.

Images should be created under the docker directory. OpenStack services should inherit from the provided openstack-base image, while supporting and infrastructure services (for example, mongodb) should inherit from base.

Services consisting of only one service should be placed in an image named the same as that service, for example, horizon. Services that consist of multiple processes generally use a base image and child images, for example, glance-base, glance-api, and glance-registry.

Jinja2 ‘blocks’ are employed throughout the Dockerfile’s to help operators customise various stages of the build (refer to Dockerfile Customisation)

Some of these blocks are free form however, there are a subset that should be common to every Dockerfile. The overall structure for a multi container service is as follows:

FROM {{ namespace }}/{{ image_prefix }}openstack-base:{{ tag }}
LABEL maintainer="{{ maintainer }}" name="{{ image_name }}" build-date="{{ build_date }}"

{% block << service >>_header %}{% endblock %}

{% import "macros.j2" as macros with context %}

<< binary specific steps >>

<< source specific steps >>

<< common steps >>

{% block << service >>_footer %}{% endblock %}
{% block footer %}{% endblock %}

Note

The generic footer block {% block footer %}{% endblock %} should not be included in base images (for example, glance-base).