The Project Repository¶
- Create your own fork of the repository
- Clone it to your projects folder
It's recommended to put all of your git checkouts in the same directory. This makes it easier to include other projects from source.
galaxy/ ├── ansible-hub-ui/ ├── galaxy-importer/ ├── galaxy_ng/ └── pulp_ansible/
cd your/preferred/projects/folder git clone email@example.com:<YOUR_USER_NAME>/galaxy_ng.git
Configuring your local code editor¶
Set your working directory to Galaxy folder
You can use your editor of choice and if you want to have the editor (ex: VsCode) to inspect the code for you might need to create a virtual environment and install the packages.
python -m venv .venv source .venv/bin/activate python -m pip install -r dev_requirements.txt python -m pip install -r docs_requirements.txt python -m pip install -r integration_requirements.txt python -m pip install -e .
Now you can for example open
code . and have VsCode to find the libraries you need.
Running the container based dev environment¶
Our containerized development environment configuration is loaded from
.compose.env script. You'll first need to configure it, and the
easiest way to do that is by copying an example script
cp .compose.env.example .compose.env
All of your local project settings can be set in your
Enable the UI (optional)¶
If you would like to develop using the UI, simply do the following:
Clone https://github.com/ansible/ansible-hub-ui to the same path where
cd your/preferred/projects/folder git clone https://github.com/ansible/ansible-hub-ui cd galaxy_ng
.compose.envfile to point to the location of the cloned UI repo. Absolute paths aren't required, but they're easier to set up. If you want to use a relative path, it has to be relative to
Complete the rest of the steps in the next section. Once everything is running the UI can be accessed at http://localhost:8002
Run the Build Steps¶
Next, run the following steps to build your local development environment:
Build the docker image
Initialize and migrate the database
Load dev data
make docker/loaddata make docker/loadtoken
You can run everything at once with
make docker/build docker/migrate docker/loaddata docker/loadtoken
Start the services¶
In foreground keeping terminal opened for watching outputs
In Background (you can close the terminal later)
./compose up -d
By default, the development environment is configured to run in
insights mode, which requires a 3rd party authentication provider. If
you're working outside of the Red Hat cloud platform, you'll want to
switch it to standalone mode by modifying your
and setting the
COMPOSE_PROFILE variable to
standalone, as shown in
the following example:
If you want to run in standalone mode while using Keycloak to provide
single sign-on with a
pre-populated LDAP server <https://github.com/rroemhild/docker-test-openldap>_
you'll want to switch it to standalone-keycloak mode by modifying your
.compose.env file, and setting the
COMPOSE_PROFILE variable to
standalone-keycloak, as shown in the following example:
Other Development Modes¶
"Insights" mode refers to running Galaxy NG as it would be run on console.redhat.com.
This option is only relevant to Red Hat employees. Community contributors should skip this.
- In your
COMPOSE_PROFILE=insightsand comment out
ANSIBLE_HUB_UI_PATH(if its set).
- Install node. Node v16+ is known to work. Older versions may work as well.
Switch to your
ansible-hub-uicheckout and run the following
npm install npm run start
The app will run on http://localhost:8002/beta/ansible/automation-hub and proxy requests for
/api/automation-hub to the api on
Keycloak mode launches an LDAP and keycloak server via docker and configures the app to authenticate using keycloak.
To run in keycloak mode set
COMPOSE_PROFILE=standalone-keycloak in your
.compose.env. You will need to initialize your Keycloak
instance before running migrations and starting the remaining services.
Start the Keycloak instance and dependencies
./compose up -d keycloak kc-postgres ldap
Bootstrap the Keycloak instance with a Realm and Client then capture the needed public key
NOTE Try again if it fails at the first run, services might not be available yet.
.compose.envfile with the public key found at the end of the playbook run
After the standard development set up steps, when you access
http://localhost:8002 it will redirect to a Keycloak Open ID Connect
flow login page, where you can login with one of the development SSO
user's credentials (the password is the username). If you want to login
with non-Keycloak users, you need to use the
Django admin console <http://localhost:5001/automation-hub/admin/>_.
If you want to login as a superuser, you can do one of two things:
Login to the
Django admin console <http://localhost:5001/automation-hub/admin/>_ with the admin user
Login to the
Keycloak instance <http://localhost:8080/>_ with admin/admin to edit the LDAP user's roles: Choose a development SSO user, select Role Mappings > Client Roles > automation-hub and add the
hubadminrole. A user is associated with the appropriate group(s) using the user_group pipeline.
Running API tests¶
Unit and integration tests can be easily run using docker compose. At the moment, there is no easy way to run pulp functional tests with docker.
For more information on tests, refer to writing tests.
Run unit all unit tests:
Run a specific test case:
make docker/test/unit TEST=.api.test_api_ui_sync_config
Integration tests can be run from the host machine or via a docker container. Before running either, the following steps have to be taken:
make docker/loadtoken make docker/load_test_data
Tests that require docker or podman won't run inside docker and will be skipped. If you need to write container tests, run the integration tests via the host machine.
Run all the integration tests:
Any set of pytest flags can be passed to this command as well:
# run any test who's name matches my_test make docker/test/integration/container FLAGS="-k my_test" # run tests marked as galaxyapi_smoke make docker/test/integration/container FLAGS="-m galaxyapi_smoke"
Via host machine¶
This requires that the
python executable in your shell be python 3, and may not work on systems synch as Mac OS where
python refers to
python2. This also requires that
virtualenv in installed on your machine.
Run all the integration tests:
Run integration tests with a specific mark:
make docker/test/integration MARK=galaxyapi_smoke
Push Container Images to local registry
To push images into the container image registry hosted by galaxy_ng (via pulp_container), you need to tag an image first to tell Docker or Podman that you want to associate the image with the registry. On a local development setup, the pulp_container runs along with the Galaxy API on port 5001.
Tag an image like this:
docker image tag <IMAGE_ID> localhost:5001/<NAME>:<VERSION>
or, to associate with a namespace:
docker image tag <IMAGE_ID> localhost:5001/<NAMESPACE>/<NAME>:<VERSION>
And then push the image and the engine will upload it to the now-associated registry:
docker push localhost:5001/testflat
Creating a set of collections for testing
When running docker environment, the project's parent directory is
mounted into container as
/app. All projects listed in
DEV_SOURCE_PATH environment variable are installed in editable mode
pip install -e) in the container. To load additional projects such as
pulp_ansible into the container from host file
system you should clone them into the parent directory relative to your
galaxy_ng working copy location.
For example you want to work on
galaxy-importer project and run
development environment with your changes made locally.
galaxy-importerto parent directory::
cd your/preferred/projects/folder git clone https://github.com/ansible/galaxy-importer cd galaxy_ng
DEV_SOURCE_PATHvariable in your
Recreate your development environment::
./compose down make docker/build docker/migrate ./compose up
The step above can be done for other Pulp plugins such as
Steps to run dev environment with specific upstream branch¶
pulp_ansibleall the repos must be located at the same directory level.
cd ~/projects/ git clone https://github.com/pulp/pulpcore git clone https://github.com/pulp/pulp_ansible git clone https://github.com/ansible/galaxy_ng # and optionally git clone https://github.com/ansible/ansible-hub-ui git clone https://github.com/ansible/galaxy_importer
Checkout to desired branches.
pulp_ansiblemain is compatible with a specific range of
pulpcoreversions. So it is recommended to checkout to a specific branch or tag following the constraints defined on pulp_ansible/requirements.txt or leave it checked out to main if you know it is compatible with the pulp_ansible branch you have. Example:
cd ~/projects/pulpcore git checkout 3.9.0
This is also possible to checkout to specific pull-requests by its
cd ~/projects/galaxy_ng cat .compose.env COMPOSE_PROFILE=standalone DEV_SOURCE_PATH='pulpcore:pulp_ansible:galaxy_ng' LOCK_REQUIREMENTS=0
DEV_SOURCE_PATH refers to the repositories you cloned locally, the order is important from the highest to the low dependency, otherwise pip will raise version conflicts.
So pulpcore is a dependency to pulp_ansible which is a dependency to galaxy_ng, this order must be respected on DEV_SOURCE_PATH variable.
LOCK_REQUIREMENTS when set to 0 it tells docker to bypass the install of pinned requirements and rely only on packages defined on
setup.pyfor each repo.
./compose buildto make those changes effective.
Run desired compose command:
Bumping The Version¶
The canonical source of truth for the 'version' is now in setup.cfg in
[bumpversion] current_version = 4.3.0.dev
To update version, it is recommended to "bump" the version instead of explicitly specifying it.
Use bump2version to increment the 'version' string wherever it is needed.
It can 'bump' the 'patch', 'minor', 'major' version components.
There are also Makefile targets for bumping versions. To do a 'patch' version bump, for example:
$ make dev/bumpversion-patch
The above command will rev the 'patch' level and update all the files that use it.
Note: Currently, the bump2version config does not git commit or git tag the changes. So after bumping the version, you need to commit the changes and tag manually.
$ git commit -v -a $ git tag $NEWVERSION
bump2version can also do this automatically if we want to enable it.
Add galaxy-importer.cfg to dev environment¶
To set your own galaxy-importer.cfg, add something like this to
RUN mkdir /etc/galaxy-importer \ && printf "[galaxy-importer]\n \ REQUIRE_V1_OR_GREATER = True\n \ LOG_LEVEL_MAIN = INFO\n" | tee /etc/galaxy-importer/galaxy-importer.cfg