Development Setup¶

The Project Repository¶

Create your own fork of the repository
Clone it to your projects folder

Note

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 git@github.com:<YOUR_USER_NAME>/galaxy_ng.git

Configuring your local code editor¶

Set your working directory to Galaxy folder

cd galaxy_ng

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 the .compose.env script. You'll first need to configure it, and the easiest way to do that is by copying an example script .compose.env.example:

cp .compose.env.example .compose.env

All of your local project settings can be set in your .compose.env file.

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 galaxy_ng is located.

cd your/preferred/projects/folder
git clone https://github.com/ansible/ansible-hub-ui
cd galaxy_ng

Set ANSIBLE_HUB_UI_PATH in your .compose.env file 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 dev/docker-compose.yml
```
ANSIBLE_HUB_UI_PATH='/your/preferred/projects/folder/ansible-hub-ui'
```
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
```
make docker/build
```
Initialize and migrate the database
```
make docker/migrate
```

Load dev data

make docker/loaddata 
make docker/loadtoken

Tip

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

./compose up

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 3^rd 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 .compose.env file, and setting the COMPOSE_PROFILE variable to standalone, as shown in the following example:

COMPOSE_PROFILE=standalone

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:

COMPOSE_PROFILE=standalone-keycloak

Other Development Modes¶

Insights¶

"Insights" mode refers to running Galaxy NG as it would be run on console.redhat.com.

Note

This option is only relevant to Red Hat employees. Community contributors should skip this.

In your .compose.env file set COMPOSE_PROFILE=insights and comment out ANSIBLE_HUB_UI_PATH (if its set).
Install node. Node v18+ is known to work. Older versions may work as well.
Switch to your ansible-hub-ui checkout 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 http://localhost:5001.

Keycloak¶

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
```
ansible-playbook ./dev/standalone-keycloak/keycloak-playbook.yaml
```
NOTE Try again if it fails at the first run, services might not be available yet.
Update your .compose.env file with the public key found at the end of the playbook run
```
PULP_SOCIAL_AUTH_KEYCLOAK_PUBLIC_KEY="keycloak-public-key"
```

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 hubadmin role. 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.

Unit tests¶

Run unit all unit tests:

make docker/test/unit

Run a specific test case:

make docker/test/unit TEST=.api.test_api_ui_sync_config

Integration tests¶

Integration tests can be run from the host machine or via a docker container. Before running either, the following steps have to be taken:

set PULP_GALAXY_REQUIRE_CONTENT_APPROVAL=true in your .compose.env.

run

make docker/loadtoken
make docker/load_test_data

Via docker¶

Note

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:

make docker/test/integration/container

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¶

Warning

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:

make docker/test/integration

Run integration tests with a specific mark:

make docker/test/integration MARK=galaxyapi_smoke

Testing data¶

Push Container Images to local registry

Info

make api/push-test-images

will push a bunch of testing images to your running system

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

Info

make api/create-test-collections

Will generate collections and populate the system

Additional Dependencies¶

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 galaxy-importer or 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.

Clone galaxy-importer to parent directory::

cd your/preferred/projects/folder
git clone https://github.com/ansible/galaxy-importer
cd galaxy_ng

Add galaxy-importer to DEV_SOURCE_PATH variable in your .compose.env file::
```
export DEV_SOURCE_PATH='galaxy_ng:galaxy-importer'
```

Recreate your development environment::

./compose down 
make docker/build docker/migrate
./compose up

Tip

The step above can be done for other Pulp plugins such as pulp_ansible or pulp_container

Steps to run dev environment with specific upstream branch¶

Clone locally galaxy_ng, pulpcore and pulp_ansible all 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_ansible main is compatible with a specific range of pulpcore versions. 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 refs/pull/id.
Edit the galaxy_ng/.compose.env file.
```
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.py for each repo.
Run ./compose build to make those changes effective.
Run desired compose command: ./compose up, ./compose run etc...

Bumping The Version¶

The canonical source of truth for the 'version' is now in setup.cfg in the bumpversion stanza:

[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.

Debugging¶

https://github.com/ansible/galaxy_ng/wiki/Debugging-with-PDB

Add galaxy-importer.cfg to dev environment¶

To set your own galaxy-importer.cfg, add something like this to /dev/Dockerfile.base:

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