An Ansible AWX operator for Kubernetes built with Operator SDK and Ansible.
This operator is meant to provide a more Kubernetes-native installation method for AWX via an AWX Custom Resource Definition (CRD).
Note that the operator is not supported by Red Hat, and is in alpha status. For now, use it at your own risk!
This Kubernetes Operator is meant to be deployed in your Kubernetes cluster(s) and can manage one or more AWX instances in any namespace.
First you need to deploy AWX Operator into your cluster:
#> kubectl apply -f https://raw.githubusercontent.com/ansible/awx-operator/devel/deploy/awx-operator.yaml
Then create a file named my-awx.yml
with the following contents:
---
apiVersion: awx.ansible.com/v1beta1
kind: AWX
metadata:
name: awx
Finally, use kubectl
to create the awx instance in your cluster:
#> kubectl apply -f my-awx.yml
After a few minutes, the new AWX instance will be deployed. One can look at the operator pod logs in order to know where the installation process is at. This can be done by running the following command: kubectl logs -f deployments/awx-operator
.
Once deployed, the AWX instance will be accessible at http://awx.mycompany.com/
(assuming your cluster has an Ingress controller configured).
By default, the admin user is admin
and the password is available in the <resourcename>-admin-password
secret. To retrieve the admin password, run kubectl get secret <resourcename>-admin-password -o jsonpath="{.data.password}" | base64 --decode
You just completed the most basic install of an AWX instance via this operator. Congratulations !
There are three variables that are customizable for the admin user account creation.
Name | Description | Default |
---|---|---|
tower_admin_user | Name of the admin user | admin |
tower_admin_email | Email of the admin user | [email protected] |
tower_admin_password_secret | Secret that contains the admin user password | Empty string |
If tower_admin_password_secret
is not provided, the operator will look for a secret named <resourcename>-admin-password
for the admin password. If it is not present, the operator will generate a password and create a Secret from it named <resourcename>-admin-password
.
To retrieve the admin password, run kubectl get secret <resourcename>-admin-password -o jsonpath="{.data.password}" | base64 --decode
The secret that is expected to be passed should be formatted as follow:
---
apiVersion: v1
kind: Secret
metadata:
name: <resourcename>-admin-password
namespace: <target namespace>
stringData:
password: mysuperlongpassword
By default, the AWX operator is not opinionated and won't force a specific ingress type on you. So, if tower_ingress_type
is not specified as part of the Custom Resource specification, it will default to none
and nothing ingress-wise will be created.
The AWX operator provides support for two kind of Ingress
to access AWX: Ingress
and Route
, To toggle between these two options, you can add the following to your AWX CR:
- Route
---
spec:
...
tower_ingress_type: Route
- Ingress
---
spec:
...
tower_ingress_type: Ingress
tower_hostname: awx.mycompany.com
- Route
The following variables are customizable to specify the TLS termination procedure when Route
is picked as an Ingress
Name | Description | Default |
---|---|---|
tower_route_host | Common name the route answers for | Empty string |
tower_route_tls_termination_mechanism | TLS Termination mechanism (Edge, Passthrough) | Edge |
tower_route_tls_secret | Secret that contains the TLS information | Empty string |
- Ingress
The following variables are customizable to specify the TLS termination procedure when Ingress
is picked as an Ingress
Name | Description | Default |
---|---|---|
tower_ingress_annotations | Ingress annotations | Empty string |
tower_ingress_tls_secret | Secret that contains the TLS information | Empty string |
In order for the AWX instance to rely on an external database, the Custom Resource needs to know about the connection details. Those connection details should be stored as a secret and either specified as tower_postgres_configuration_secret
at the CR spec level, or simply be present on the namespace under the name <resourcename>-postgres-configuration
.
The secret should be formatted as follows:
---
apiVersion: v1
kind: Secret
metadata:
name: <resourcename>-postgres-configuration
namespace: <target namespace>
stringData:
host: <external ip or url resolvable by the cluster>
port: <external port, this usually defaults to 5432>
database: <desired database name>
username: <username to connect as>
password: <password to connect with>
type: Opaque
If you don't have access to an external PostgreSQL service, the AWX operator can deploy one for you along side the AWX instance itself.
The following variables are customizable for the managed PostgreSQL service
Name | Description | Default |
---|---|---|
tower_postgres_image | Path of the image to pull | postgres:12 |
tower_postgres_resource_requirements | PostgreSQL container resource requirements | requests: {storage: 8Gi} |
tower_postgres_storage_class | PostgreSQL PV storage class | Empty string |
tower_postgres_data_path | PostgreSQL data path | /var/lib/postgresql/data/pgdata |
Example of customization could be:
---
spec:
...
tower_postgres_resource_requirements:
requests:
memory: 2Gi
storage: 8Gi
limits:
memory: 4Gi
storage: 50Gi
tower_postgres_storage_class: fast-ssd
Note: If tower_postgres_storage_class
is not defined, Postgres will store it's data on a volume using the default storage class for your cluster.
There are two variables that are customizable for awx the image management.
Name | Description | Default |
---|---|---|
tower_image | Path of the image to pull | ansible/awx:15.0.0 |
tower_image_pull_policy | The pull policy to adopt | IfNotPresent |
Example of customization could be:
---
spec:
...
tower_image: myorg/my-custom-awx
tower_image_pull_policy: Always
Depending on the type of tasks that you'll be running, you may find that you need the task pod to run as privileged
. This can open yourself up to a variety of security concerns, so you should be aware (and verify that you have the privileges) to do this if necessary. In order to toggle this feature, you can add the following to your custom resource:
---
spec:
...
tower_task_privileged: true
If you are attempting to do this on an OpenShift cluster, you will need to grant the awx
ServiceAccount the privileged
SCC, which can be done with:
#> oc adm policy add-scc-to-user privileged -z awx
Again, this is the most relaxed SCC that is provided by OpenShift, so be sure to familiarize yourself with the security concerns that accompany this action.
The resource requirements for both, the task and the web containers are configurable - both the lower end (requests) and the upper end (limits).
Name | Description | Default |
---|---|---|
tower_web_resource_requirements | Web container resource requirements | requests: {cpu: 1000m, memory: 2Gi} |
tower_task_resource_requirements | Task container resource requirements | requests: {cpu: 500m, memory: 1Gi} |
Example of customization could be:
---
spec:
...
tower_web_resource_requirements:
requests:
cpu: 1000m
memory: 2Gi
limits:
cpu: 2000m
memory: 4Gi
tower_task_resource_requirements:
requests:
cpu: 500m
memory: 1Gi
limits:
cpu: 1000m
memory: 2Gi
This Operator includes a Molecule-based test environment, which can be executed standalone in Docker (e.g. in CI or in a single Docker container anywhere), or inside any kind of Kubernetes cluster (e.g. Minikube).
You need to make sure you have Molecule installed before running the following commands. You can install Molecule with:
#> pip install 'molecule[docker]'
Running molecule test
sets up a clean environment, builds the operator, runs all configured tests on an example operator instance, then tears down the environment (at least in the case of Docker).
If you want to actively develop the operator, use molecule converge
, which does everything but tear down the environment at the end.
#> molecule test -s test-local
This environment is meant for headless testing (e.g. in a CI environment, or when making smaller changes which don't need to be verified through a web interface). It is difficult to test things like AWX's web UI or to connect other applications on your local machine to the services running inside the cluster, since it is inside a Docker container with no static IP address.
#> minikube start --memory 8g --cpus 4
#> minikube addons enable ingress
#> molecule test -s test-minikube
Minikube is a more full-featured test environment running inside a full VM on your computer, with an assigned IP address. This makes it easier to test things like NodePort services and Ingress from outside the Kubernetes cluster (e.g. in a browser on your computer).
Once the operator is deployed, you can visit the AWX UI in your browser by following these steps:
- Make sure you have an entry like
IP_ADDRESS example-awx.test
in your/etc/hosts
file. (Get the IP address withminikube ip
.) - Visit
http://example-awx.test/
in your browser. (Default admin login istest
/changeme
.)
Alternatively, you can also update the service awx-service
in your namespace to use the type NodePort
and use following command to get the URL to access your AWX instance:
#> minikube service <serviceName> -n <namespaceName> --url
There are a few moving parts to this project:
- The Docker image which powers AWX Operator.
- The
awx-operator.yaml
Kubernetes manifest file which initially deploys the Operator into a cluster.
Each of these must be appropriately built in preparation for a new tag:
Run the following command inside this directory:
#> operator-sdk build quay.io/ansible/awx-operator:$VERSION
Then push the generated image to Docker Hub:
#> docker push quay.io/ansible/awx-operator:$VERSION
Update the awx-operator version:
ansible/group_vars/all
Once the version has been updated, run from the root of the repo:
#> ansible-playbook ansible/chain-operator-files.yml
After it is built, test it on a local cluster:
#> minikube start --memory 6g --cpus 4
#> minikube addons enable ingress
#> kubectl apply -f deploy/awx-operator.yaml
#> kubectl create namespace example-awx
#> kubectl apply -f deploy/crds/awx_v1beta1_cr.yaml
#> <test everything>
#> minikube delete
If everything works, commit the updated version, then tag a new repository release with the same tag as the Docker image pushed earlier.
This operator was originally built in 2019 by Jeff Geerling and is now maintained by the Ansible Team