Deploying your private Galaxy NG a.k.a. upstream version of Ansible Automation Hub.
Note that the containerized implementation of Galaxy NG is not supported at this time. See the official installation guide for supported procedure.
All information on this page is for development, testing and study purposes only.
- Deploy on Docker (Official Development Setup)
- Deploy on Docker (All-in-One Container)
- Deploy on Kubernetes (All-in-One Container)
- Deploy on Kubernetes (Pulp Operator)
- Configuration and Usage
- Use with AWX
Official guide for Development Setup provides the procedure to run Galaxy NG on Docker.
Development Setup · ansible/galaxy_ng Wiki
You can control the version of Galaxy NG by using Tags in cloned local Git repository. It takes some time to build the image, but it's not that complicated and it's a good way to try out Galaxy NG.
Pulp Project provides an all-in-one container image that contains all the necessary elements. One of the easiest ways to get a working Galaxy NG is to run it on Docker.
- Pulp in One Container | software repository management
- pulp/pulp-oci-images: Containerfiles and other assets for building Pulp 3 OCI images
Although not documented, a container image with Galaxy NG preinstalled and its source Containerfile
are also available.
- pulp/pulp-galaxy-ng - Docker Image | Docker Hub
- pulp-oci-images/pulp_galaxy_ng at latest · pulp/pulp-oci-images
There are only three steps to make this work.
First, prepare a directory and a configuration file. You can replace the hostname as you like.
mkdir settings pulp_storage pgsql containers
cat <<EOF > settings/settings.py
CONTENT_ORIGIN='http://$(hostname):8080'
ANSIBLE_API_HOSTNAME='http://$(hostname):8080'
ANSIBLE_CONTENT_HOSTNAME='http://$(hostname):8080/pulp/content'
TOKEN_AUTH_DISABLED=True
EOF
Then invoke docker run
.
docker run --detach \
--publish 8080:80 \
--name pulp \
--volume "$(pwd)/settings":/etc/pulp \
--volume "$(pwd)/pulp_storage":/var/lib/pulp \
--volume "$(pwd)/pgsql":/var/lib/pgsql \
--volume "$(pwd)/containers":/var/lib/containers \
--device /dev/fuse \
pulp/pulp-galaxy-ng:latest
Once it has started, load the initial configuration file.
DATA_FIXTURE_URL="https://raw.githubusercontent.com/ansible/galaxy_ng/master/dev/automation-hub/initial_data.json"
curl $DATA_FIXTURE_URL | docker exec -i pulp bash -c "cat > /tmp/initial_data.json"
docker exec pulp bash -c "/usr/local/bin/pulpcore-manager loaddata /tmp/initial_data.json"
Now your own Galaxy NG is available at http://$(hostname):8080/
. You can log in to the GUI by user admin
with password admin
.
In this step, we will run the above All-in-One container on Kubernetes.
Generate a Self-Signed Certificate. Note that IP address can't be specified.
GALAXY_HOST="galaxy.example.com"
openssl req -x509 -nodes -days 3650 -newkey rsa:2048 -out ./galaxy/all-in-one/tls.crt -keyout ./galaxy/all-in-one/tls.key -subj "/CN=${GALAXY_HOST}/O=${GALAXY_HOST}" -addext "subjectAltName = DNS:${GALAXY_HOST}"
Modify hosts
and host
in all-in-one/ingress.yaml
.
...
- hosts:
- galaxy.example.com 👈👈👈
secretName: galaxy-secret-tls
rules:
- host: galaxy.example.com 👈👈👈
...
Modify FQDNs in all-in-one/configmap.yaml
.
...
data:
settings.py: |-
CONTENT_ORIGIN='https://galaxy.example.com' 👈👈👈
ANSIBLE_API_HOSTNAME='https://galaxy.example.com' 👈👈👈
ANSIBLE_CONTENT_HOSTNAME='https://galaxy.example.com/pulp/content' 👈👈👈
TOKEN_AUTH_DISABLED=True
Prepare directories for Persistent Volumes defined in all-in-one/pv.yaml
.
sudo mkdir -p /data/galaxy
Deploy Galaxy NG.
kubectl apply -k galaxy/all-in-one
Required resources has been deployed in galaxy
namespace.
$ kubectl -n galaxy get all
NAME READY STATUS RESTARTS AGE
pod/galaxy-78df96fc64-l7tbq 1/1 Running 0 53s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/galaxy-service ClusterIP 10.43.201.53 <none> 80/TCP 6m14s
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/galaxy 1/1 1 1 53s
NAME DESIRED CURRENT READY AGE
replicaset.apps/galaxy-78df96fc64 1 1 1 53s
Once it has started, load the initial configuration file.
POD_NAME=$(kubectl -n galaxy get pod -l app=galaxy -o name)
DATA_FIXTURE_URL="https://raw.githubusercontent.com/ansible/galaxy_ng/master/dev/automation-hub/initial_data.json"
curl $DATA_FIXTURE_URL | kubectl -n galaxy exec -i $POD_NAME -- bash -c "cat > /tmp/initial_data.json"
kubectl -n galaxy exec -i $POD_NAME -- bash -c "/usr/local/bin/pulpcore-manager loaddata /tmp/initial_data.json"
Now Galaxy NG is available at https://galaxy.example.com/
or the hostname you specified. You can log in to the GUI by user admin
with password admin
.
There is a Kubernetes Operator for Pulp 3 named Pulp Operator.
This project is still under active development and there is no support, however, at least the code to create a new instance seems to be implemented. In this procedure, we use Pulp Operator 0.13.0
If you use Traefik which is K3s' Ingress controller as completely default, the Pod may not be able to get the client's IP address (see k3s-io/k3s#2997 for details). In the current implementation of Pulp, this causes problems with the web UI being unreachable.
For this reason, you should fix the Traefik configuration. For a single node like doing in this repository, reconfiguring your Traefik by creating YAML file is the easy way.
sudo tee /var/lib/rancher/k3s/server/manifests/traefik-config.yaml <<EOF
---
apiVersion: helm.cattle.io/v1
kind: HelmChartConfig
metadata:
name: traefik
namespace: kube-system
spec:
valuesContent: |-
hostNetwork: true
EOF
Then wait until your traefik
by the following command is 1/1
READY
.
kubectl -n kube-system get deployment traefik
Now your client's IP address can be passed correctly through X-Forwarded-For
and X-Real-Ip
headers.
Install specified version of Pulp Operator.
cd ~
git clone https://github.com/pulp/pulp-operator.git
cd pulp-operator
git checkout 0.13.0
Export the name of the namespace where you want to deploy Pulp Operator as the environment variable NAMESPACE
and run make deploy
. The default namespace is pulp-operator-system
.
export NAMESPACE=galaxy
make deploy
The Pulp Operator will be deployed to the namespace you specified.
$ kubectl -n galaxy get all
NAME READY STATUS RESTARTS AGE
pod/pulp-operator-controller-manager-9b8644f46-rg2rl 2/2 Running 0 21s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/pulp-operator-controller-manager-metrics-service ClusterIP 10.43.20.233 <none> 8443/TCP 21s
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/pulp-operator-controller-manager 1/1 1 1 21s
NAME DESIRED CURRENT READY AGE
replicaset.apps/pulp-operator-controller-manager-9b8644f46 1 1 1 21s
Clone this repository and change directory.
cd ~
git clone https://github.com/kurokobo/awx-on-k3s.git
cd awx-on-k3s/galaxy
Generate a Self-Signed Certificate and key pair. Note that IP address can't be specified.
GALAXY_HOST="galaxy.example.com"
openssl req -x509 -nodes -days 3650 -newkey rsa:2048 -out ./pulp/tls.crt -keyout ./pulp/tls.key -subj "/CN=${GALAXY_HOST}/O=${GALAXY_HOST}" -addext "subjectAltName = DNS:${GALAXY_HOST}"
Modify hostname
in pulp/galaxy.yaml
.
...
spec:
...
ingress_type: ingress
ingress_tls_secret: galaxy-secret-tls
hostname: galaxy.example.com 👈👈👈
...
Modify two password
s in pulp/kustomization.yaml
.
...
- name: galaxy-postgres-configuration
type: Opaque
literals:
- host=galaxy-postgres
- port=5432
- database=galaxy
- username=galaxy
- password=Galaxy123! 👈👈👈
- type=managed
- name: galaxy-admin-password
type: Opaque
literals:
- password=Galaxy123! 👈👈👈
...
Prepare directories for Persistent Volumes defined in pulp/pv.yaml
.
sudo mkdir -p /data/galaxy/postgres-13
sudo mkdir -p /data/galaxy/redis
sudo mkdir -p /data/galaxy/pulp
Deploy Galaxy NG.
kubectl apply -k pulp
To monitor the progress of the deployment, check the logs of deployments/awx-operator-controller-manager
:
kubectl -n galaxy logs -f deployments/pulp-operator-controller-manager -c pulp-manager
When the deployment completes successfully, the logs end with:
$ kubectl -n galaxy logs -f deployments/pulp-operator-controller-manager -c pulp-manager
...
----- Ansible Task Status Event StdOut (pulp.pulpproject.org/v1beta1, Kind=Pulp, galaxy/galaxy) -----
PLAY RECAP *********************************************************************
localhost : ok=86 changed=0 unreachable=0 failed=0 skipped=69 rescued=0 ignored=0
Required objects has been deployed next to Pulp Operator in galaxy
namespace.
$ kubectl -n galaxy get pulp,all,ingress,secrets
NAME AGE
pulp.pulp.pulpproject.org/galaxy 4m48s
NAME READY STATUS RESTARTS AGE
pod/pulp-operator-controller-manager-74d4f59ddb-mtwzq 2/2 Running 0 5m4s
pod/galaxy-postgres-13-0 1/1 Running 0 4m31s
pod/galaxy-redis-6f79b47f66-4dg75 1/1 Running 0 4m5s
pod/galaxy-content-57fd45b97c-mxkdx 1/1 Running 0 3m39s
pod/galaxy-worker-6fcbd5ff47-b2h6v 1/1 Running 0 3m37s
pod/galaxy-api-5cf69c85d-vc6z8 1/1 Running 0 3m43s
pod/galaxy-web-5f7bc488fb-69mbk 1/1 Running 0 4m22s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/pulp-operator-controller-manager-metrics-service ClusterIP 10.43.145.99 <none> 8443/TCP 5m4s
service/galaxy-postgres-13 ClusterIP None <none> 5432/TCP 4m31s
service/galaxy-web-svc ClusterIP 10.43.17.192 <none> 24880/TCP 4m21s
service/galaxy-redis-svc ClusterIP 10.43.58.131 <none> 6379/TCP 4m7s
service/galaxy-api-svc ClusterIP 10.43.106.139 <none> 24817/TCP 3m52s
service/galaxy-content-svc ClusterIP 10.43.154.10 <none> 24816/TCP 3m41s
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/pulp-operator-controller-manager 1/1 1 1 5m4s
deployment.apps/galaxy-redis 1/1 1 1 4m5s
deployment.apps/galaxy-content 1/1 1 1 3m39s
deployment.apps/galaxy-worker 1/1 1 1 3m37s
deployment.apps/galaxy-api 1/1 1 1 3m43s
deployment.apps/galaxy-web 1/1 1 1 4m22s
NAME DESIRED CURRENT READY AGE
replicaset.apps/pulp-operator-controller-manager-74d4f59ddb 1 1 1 5m4s
replicaset.apps/galaxy-redis-6f79b47f66 1 1 1 4m5s
replicaset.apps/galaxy-content-57fd45b97c 1 1 1 3m39s
replicaset.apps/galaxy-worker-6fcbd5ff47 1 1 1 3m37s
replicaset.apps/galaxy-api-5cf69c85d 1 1 1 3m43s
replicaset.apps/galaxy-web-5f7bc488fb 1 1 1 4m22s
NAME READY AGE
statefulset.apps/galaxy-postgres-13 1/1 4m31s
NAME CLASS HOSTS ADDRESS PORTS AGE
ingress.networking.k8s.io/galaxy-ingress <none> galaxy.example.com 192.168.0.100 80, 443 4m19s
NAME TYPE DATA AGE
secret/default-token-lhlds kubernetes.io/service-account-token 3 5m4s
secret/pulp-operator-sa-token-45zqf kubernetes.io/service-account-token 3 5m4s
secret/galaxy-admin-password Opaque 1 4m48s
secret/galaxy-postgres-configuration Opaque 6 4m48s
secret/galaxy-secret-tls kubernetes.io/tls 2 4m48s
secret/redhat-operators-pull-secret Opaque 1 4m40s
secret/galaxy-server Opaque 1 4m3s
secret/galaxy-db-fields-encryption Opaque 1 3m54s
secret/galaxy-container-auth Opaque 2 3m47s
Now your AWX is available at https://galaxy.example.com/
or the hostname you specified. You can log in to the GUI by user admin
with password you specified in pulp/kustomization.yaml
.
Basic configuration and usage of Galaxy NG.
Create a list of Collections to be synchronized as YAML file.
---
collections:
- name: community.general
source: https://galaxy.ansible.com
version: ">=3.2.0"
- name: community.kubernetes
source: https://galaxy.ansible.com
version: "2.0.0"
- name: community.vmware
source: https://galaxy.ansible.com
version: ">=1.10.0,<1.12.0"
- name: awx.awx
source: https://galaxy.ansible.com
version: ">=19.0.0"
- name: ansible.utils
source: https://galaxy.ansible.com
version: ">=2.1.0"
In Galaxy NG, open Collections
> Repository Management
> Remote
and click Configure
on community
.
Select your YAML file in YAML requirements
and Save
, then Sync
and wait to complete.
Create a new minimal collection with a minimal plugin, minimal module, and minimal role.
# Create skeleton collection
ansible-galaxy collection init demo.collection
# Create meta file
mkdir -p demo/collection/meta
cat <<EOF > demo/collection/meta/runtime.yml
---
requires_ansible: ">=2.9"
EOF
# Create new Plugin
mkdir -p demo/collection/plugins/vars
cat <<EOF > demo/collection/plugins/vars/sample_vars.py
DOCUMENTATION = '''
---
vars: sample_vars
short_description: Add a fixed variable named sample_var
version_added: "1.0.0"
description: Just add a fixed variable with name sample_var.
'''
from ansible.plugins.vars import BaseVarsPlugin
class VarsModule(BaseVarsPlugin):
def get_vars(self, loader, path, entities):
return {"sample_var": "This is sample variable"}
EOF
# Create new Module
mkdir -p demo/collection/plugins/modules
cat <<EOF > demo/collection/plugins/modules/sample_module.py
DOCUMENTATION = '''
---
module: sample_module
short_description: This is my test module
version_added: "1.0.0"
description: This is my longer description explaining my test module.
'''
from ansible.module_utils.basic import AnsibleModule
if __name__ == '__main__':
result = dict(changed=False, message='Hello from Module')
module = AnsibleModule(argument_spec={})
module.exit_json(**result)
EOF
# Create new Role
cd demo/collection/roles
ansible-galaxy init sample_role
cat <<EOF > sample_role/tasks/main.yml
---
- name: Hello
debug:
msg: "World"
EOF
# Build tarball
cd ../
ansible-galaxy collection build
Then create demo
namespace on Galaxy NG, and publish your collection.
Note that you can get appropriate URL for --server
from Collections
> Namespaces
> View collections
> CLI Configuration
per collections. Your token is available at Collections
> API token management
> Load token
.
ansible-galaxy collection publish \
demo-collection-1.0.0.tar.gz \
--server https://galaxy.example.com/api/galaxy/content/inbound-demo/ \
--token d926e******************************3e996 \
-c
Once the command succeeded, your collection is stayed at staging
distribution. Approval by super user on Collections
> Approval
page is required to move your collection to published
distribution.
Optionally, this approval process can be disabled by adding galaxy_require_content_approval: "False"
in your settings.py
.
Modify your ansible.cfg
to specify which Galaxy Instance will be used in which order. Note that you can get appropriate configuration from Collections
> Repository Management
> Local
> CLI configuration
per distributions. Your token is available at Collections
> API Token
.
[galaxy]
server_list = published_repo, community_repo
[galaxy_server.published_repo]
url=https://galaxy.example.com/api/galaxy/content/published/
token=d926e******************************3e996
[galaxy_server.community_repo]
url=https://galaxy.example.com/api/galaxy/content/community/
token=d926e******************************3e996
Then simply collection install
command can be used to install collections from your Galaxy NG.
ansible-galaxy collection install community.vmware -c
ansible-galaxy collection install demo.collection -c
You can test your collection by invoking minimal playbook.
cat <<EOF > site.yml
- hosts: localhost
roles:
- demo.collection.sample_role
tasks:
- demo.collection.sample_module:
register: result
- debug:
var: result
EOF
ansible-playbook site.yml
Add galaxy.example.com
as an Insecure Registry.
sudo tee /etc/docker/daemon.json <<EOF
{
"insecure-registries" : ["galaxy.example.com"]
}
EOF
sudo systemctl restart docker
Then simply login
, tag
and push
.
docker login galaxy.example.com
docker tag registry.example.com/ansible/ee:2.12-custom galaxy.example.com/demo/ee:2.12-custom
docker push galaxy.example.com/demo/ee:2.12-custom
Use your Galaxy NG with AWX.
To use your Collections on your Galaxy NG through AWX, some tasks are required before creating Project.
- Store your Token and URL for specific Organization in AWX
- Add credential with type
Ansible Galaxy/Automation Hub API Token
with your Token and Galaxy Server URL. - You can get appropriate URL from
Collections
>Repository Management
>Local
>CLI configuration
per distributions on Galaxy NG. - Your token is available at
Collections
>API Token
on Galaxy NG.
- Add credential with type
- Enable your credential in Organization
- In
Edit
screen forOrganization
that will use your Galaxy NG, enable your credential inGalaxy Credentials
. - You can change the order of credentials to set precedence for the sync and lookup of the content.
- In
- Ignore SSL Certificate Verification
- Enable
Ignore Ansible Galaxy SSL Certificate Verification
inSettings
>Jobs
>Jobs settings
- Enable
Then create files to test collection.
mkdir collection-demo
cd collection-demo
mkdir collections
cat <<EOF > collections/requirements.yml
---
collections:
- name: demo.collection
# If you want to ignore search order and get collections from a specific Galaxy Server,
# you can specify the source URL here
#source: https://galaxy.example.com/api/galaxy/content/published/
EOF
cat <<EOF > site.yml
---
- hosts: localhost
roles:
- demo.collection.sample_role
tasks:
- demo.collection.sample_module:
register: result
- debug:
var: result
EOF
Push files in collection-demo
to your SCM, and create new Project in AWX in standard way. Once Sync has been invoked, your collection will be installed through your Galaxy NG.
To use your Execution Environment on your Galaxy NG through AWX, Kubernetes have to be able to pull images from your Galaxy NG.
If the endpoint of the Galaxy NG you created is HTTPS with a Self-Signed Certificate, you need to disable SSL validation for the registry.
To achieve this, create a registries.yaml
, and then restart K3s.
sudo tee /etc/rancher/k3s/registries.yaml <<EOF
configs:
galaxy.example.com:
tls:
insecure_skip_verify: true
EOF
# The K3s service can be safely restarted without affecting the running resources
sudo systemctl restart k3s
If this is successfully applied, you can check the applied configuration in the config.registry
section of the following command.
sudo /usr/local/bin/crictl info
# With jq
sudo /usr/local/bin/crictl info | jq .config.registry
Now you can use Execution Environment on Galaxy NG through AWX as following.
- Push your Execution Environment to your Galaxy NG (as described above)
- Create Credential with
Container Registry
type on AWX for your Galaxy NG - Register new Execution Environment on AWX
- Specify it as Execution Environment for the Job Template, Project Default, or Global Default.
Once you start the Job Template, imagePullSecrets
will be created from Credentials and assigned to the Pod, the image will be pulled, and the playbook will run on the Execution Environment.