This document provides a guide for installing AWX.
If you have not already done so, you will need to clone, or create a local copy, of the AWX repo. For more on how to clone the repo, view git clone help.
Once you have a local copy, run commands within the root of the project tree.
You can optionally install the AWX branding assets from the awx-logos repo. Prior to installing, please review and agree to the trademark guidelines.
To install the assets, clone the awx-logos
repo so that it is next to your awx
clone. As you progress through the installation steps, you'll be setting variables in the inventory file. To include the assets in the build, set awx_official=true
.
Before you can run a deployment, you'll need the following installed in your local environment:
- Ansible Requires Version 2.4+
- Docker
- docker-py Python module
- GNU Make
- Git Requires Version 1.8.4+
- Node 8.x LTS version
- NPM 6.x LTS
The system that runs the AWX service will need to satisfy the following requirements
- At leasts 4GB of memory
- At least 2 cpu cores
- At least 20GB of space
- Running Docker, Openshift, or Kubernetes
- If you choose to use an external PostgreSQL database, please note that the minimum version is 9.6+.
TODO add tunable bits
We currently support running AWX as a containerized application using Docker images deployed to either an OpenShift cluster or docker-compose. The remainder of this document will walk you through the process of building the images, and deploying them to either platform.
The installer directory contains an inventory file, and a playbook, install.yml. You'll begin by setting variables in the inventory file according to the platform you wish to use, and then you'll start the image build and deployment process by running the playbook.
In the sections below, you'll find deployment details and instructions for each platform:
When installing AWX you have the option of building your own images or using the images provided on DockerHub (see awx_web and awx_task)
This is controlled by the following variables in the inventory
file
dockerhub_base=ansible
dockerhub_version=latest
If these variables are present then all deployments will use these hosted images. If the variables are not present then the images will be built during the install.
dockerhub_base
The base location on DockerHub where the images are hosted (by default this pulls container images named
ansible/awx_web:tag
andansible/awx_task:tag
)
dockerhub_version
Multiple versions are provided.
latest
always pulls the most recent. You may also select version numbers at different granularities: 1, 1.0, 1.0.1, 1.0.0.123
To complete a deployment to OpenShift, you will obviously need access to an OpenShift cluster. For demo and testing purposes, you can use Minishift to create a single node cluster running inside a virtual machine.
You will also need to have the oc
command in your PATH. The install.yml
playbook will call out to oc
when logging into, and creating objects on the cluster.
The default resource requests per-deployment requires:
Memory: 6GB CPU: 3 cores
This can be tuned by overriding the variables found in /installer/roles/kubernetes/defaults/main.yml. Special care should be taken when doing this as undersized instances will experience crashes and resource exhaustion.
For more detail on how resource requests are formed see: https://docs.openshift.com/container-platform/latest/dev_guide/compute_resources.html#dev-compute-resources
Before starting the build process, review the inventory file, and uncomment and provide values for the following variables found in the [all:vars]
section:
openshift_host
IP address or hostname of the OpenShift cluster. If you're using Minishift, this will be the value returned by
minishift ip
.
openshift_skip_tls_verify
Boolean. Set to True if using self-signed certs.
openshift_project
Name of the OpenShift project that will be created, and used as the namespace for the AWX app. Defaults to awx.
openshift_user
Username of the OpenShift user that will create the project, and deploy the application. Defaults to developer.
openshift_pg_emptydir
Boolean. Set to True to use an emptyDir volume when deploying the PostgreSQL pod. Note: This should only be used for demo and testing purposes.
docker_registry
IP address and port, or URL, for accessing a registry that the OpenShift cluster can access. Defaults to 172.30.1.1:5000, the internal registry delivered with Minishift. This is not needed if you are using official hosted images.
docker_registry_repository
Namespace to use when pushing and pulling images to and from the registry. Generally this will match the project name. It defaults to awx. This is not needed if you are using official hosted images.
docker_registry_username
Username of the user that will push images to the registry. Will generally match the openshift_user value. Defaults to developer. This is not needed if you are using official hosted images.
Install Minishift by following the installation guide.
The recommended minimum resources for your Minishift VM:
$ minishift start --cpus=4 --memory=8GB
The Minishift VM contains a Docker daemon, which you can use to build the AWX images. This is generally the approach you should take, and we recommend doing so. To use this instance, run the following command to setup your environment:
# Set DOCKER environment variable to point to the Minishift VM
$ eval $(minishift docker-env)
Note
If you choose to not use the Docker instance running inside the VM, and build the images externally, you will have to enable the OpenShift cluster to access the images. This involves pushing the images to an external Docker registry, and granting the cluster access to it, or exposing the internal registry, and pushing the images into it.
By default, AWX will deploy a PostgreSQL pod inside of your cluster. You will need to create a Persistent Volume Claim which is named postgresql
by default, and can be overridden by setting the openshift_pg_pvc_name
variable. For testing and demo purposes, you may set openshift_pg_emptydir=yes
.
If you wish to use an external database, in the inventory file, set the value of pg_hostname
, and update pg_username
, pg_password
, pg_database
, and pg_port
with the connection information. When setting pg_hostname
the installer will assume you have configured the database in that location and will not launch the postgresql pod.
To start the build, you will pass two extra variables on the command line. The first is openshift_password, which is the password for the openshift_user, and the second is docker_registry_password, which is the password associated with docker_registry_username.
If you're using the OpenShift internal registry, then you'll pass an access token for the docker_registry_password value, rather than a password. The oc whoami -t
command will generate the required token, as long as you're logged into the cluster via oc cluster login
.
To start the build and deployment, run the following (docker_registry_password is optional if using official images):
# Start the build and deployment
$ ansible-playbook -i inventory install.yml -e openshift_password=developer -e docker_registry_password=$(oc whoami -t)
After the playbook run completes, check the status of the deployment by running oc get pods
:
# View the running pods
$ oc get pods
NAME READY STATUS RESTARTS AGE
awx-3886581826-5mv0l 4/4 Running 0 8s
postgresql-1-l85fh 1/1 Running 0 20m
In the above example, the name of the AWX pod is awx-3886581826-5mv0l
. Before accessing the AWX web interface, setup tasks and database migrations need to complete. These tasks are running in the awx_task
container inside the AWX pod. To monitor their status, tail the container's STDOUT by running the following command, replacing the AWX pod name with the pod name from your environment:
# Follow the awx_task log output
$ oc logs -f awx-3886581826-5mv0l -c awx-celery
You will see the following indicating that database migrations are running:
Using /etc/ansible/ansible.cfg as config file
127.0.0.1 | SUCCESS => {
"changed": false,
"db": "awx"
}
Operations to perform:
Synchronize unmigrated apps: solo, api, staticfiles, messages, channels, django_extensions, ui, rest_framework, polymorphic
Apply all migrations: sso, taggit, sessions, sites, kombu_transport_django, social_auth, contenttypes, auth, conf, main
Synchronizing apps without migrations:
Creating tables...
Running deferred SQL...
Installing custom SQL...
Running migrations:
Rendering model states... DONE
Applying contenttypes.0001_initial... OK
Applying contenttypes.0002_remove_content_type_name... OK
Applying auth.0001_initial... OK
Applying auth.0002_alter_permission_name_max_length... OK
Applying auth.0003_alter_user_email_max_length... OK
Applying auth.0004_alter_user_username_opts... OK
Applying auth.0005_alter_user_last_login_null... OK
Applying auth.0006_require_contenttypes_0002... OK
Applying taggit.0001_initial... OK
Applying taggit.0002_auto_20150616_2121... OK
...
When you see output similar to the following, you'll know that database migrations have completed, and you can access the web interface:
Python 2.7.5 (default, Nov 6 2016, 00:28:07)
[GCC 4.8.5 20150623 (Red Hat 4.8.5-11)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> <User: admin>
>>> Default organization added.
Demo Credential, Inventory, and Job Template added.
Successfully registered instance awx-3886581826-5mv0l
(changed: True)
Creating instance group tower
Added instance awx-3886581826-5mv0l to tower
Once database migrations complete, the web interface will be accessible.
The AWX web interface is running in the AWX pod, behind the awx-web-svc
service. To view the service, and its port value, run the following command:
# View available services
$ oc get services
NAME CLUSTER-IP EXTERNAL-IP PORT(S) AGE
awx-web-svc 172.30.111.74 <nodes> 8052:30083/TCP 37m
postgresql 172.30.102.9 <none> 5432/TCP 38m
The deployment process creates a route, awx-web-svc
, to expose the service. How the ingres is actually created will vary depending on your environment, and how the cluster is configured. You can view the route, and the external IP address and hostname assigned to it, by running the following command:
# View available routes
$ oc get routes
NAME HOST/PORT PATH SERVICES PORT TERMINATION WILDCARD
awx-web-svc awx-web-svc-awx.192.168.64.2.nip.io awx-web-svc http edge/Allow None
The above example is taken from a Minishift instance. From a web browser, use https
to access the HOST/PORT
value from your environment. Using the above example, the URL to access the server would be https://awx-web-svc-awx.192.168.64.2.nip.io.
Once you access the AWX server, you will be prompted with a login dialog. The default administrator username is admin
, and the password is password
.
A Kubernetes deployment will require you to have access to a Kubernetes cluster as well as the following tools:
The installation program will reference kubectl
directly. helm
is only necessary if you are letting the installer configure PostgreSQL for you.
The default resource requests per-pod requires:
Memory: 6GB CPU: 3 cores
This can be tuned by overriding the variables found in /installer/roles/kubernetes/defaults/main.yml. Special care should be taken when doing this as undersized instances will experience crashes and resource exhaustion.
For more detail on how resource requests are formed see: https://kubernetes.io/docs/concepts/configuration/manage-compute-resources-container/
Before starting the build process, review the inventory file, and uncomment and provide values for the following variables found in the [all:vars]
section uncommenting when necessary. Make sure the openshift and standalone docker sections are commented out:
kubernetes_context
Prior to running the installer, make sure you've configured the context for the cluster you'll be installing to. This is how the installer knows which cluster to connect to and what authentication to use
kubernetes_namespace
Name of the Kubernetes namespace where the AWX resources will be installed. This will be created if it doesn't exist
docker_registry_
These settings should be used if building your own base images. You'll need access to an external registry and are responsible for making sure your kube cluster can talk to it and use it. If these are undefined and the dockerhub_ configuration settings are uncommented then the images will be pulled from dockerhub instead
If you want the AWX installer to manage creating the database pod (rather than installing and configuring postgres on your own). Then you will need to have a working helm
installation, you can find details here: https://docs.helm.sh/using_helm/#quickstart-guide.
Newer Kubernetes clusters with RBAC enabled will need to make sure a service account is created, make sure to follow the instructions here https://docs.helm.sh/using_helm/#role-based-access-control
After making changes to the inventory
file use ansible-playbook
to begin the install
$ ansible-playbook -i inventory install.yml
After the playbook run completes, check the status of the deployment by running kubectl get pods --namespace awx
(replace awx with the namespace you used):
# View the running pods, it may take a few minutes for everything to be marked in the Running state
$ kubectl get pods --namespace awx
NAME READY STATUS RESTARTS AGE
awx-2558692395-2r8ss 4/4 Running 0 29s
awx-postgresql-355348841-kltkn 1/1 Running 0 1m
The AWX web interface is running in the AWX pod behind the awx-web-svc
service:
# View available services
$ kubectl get svc --namespace awx
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
awx-postgresql ClusterIP 10.7.250.208 <none> 5432/TCP 2m
awx-web-svc NodePort 10.7.241.35 <none> 80:30177/TCP 1m
The deployment process creates an Ingress
named awx-web-svc
also. Some kubernetes cloud providers will automatically handle routing configuration when an Ingress is created others may require that you more explicitly configure it. You can see what kubernetes knows about things with:
kubectl get ing --namespace awx
NAME HOSTS ADDRESS PORTS AGE
awx-web-svc * 35.227.x.y 80 3m
If your provider is able to allocate an IP Address from the Ingress controller then you can navigate to the address and access the AWX interface. For some providers it can take a few minutes to allocate and make this accessible. For other providers it may require you to manually intervene.
Unlike Openshift's Route
the Kubernetes Ingress
doesn't yet handle SSL termination. As such the default configuration will only expose AWX through HTTP on port 80. You are responsible for configuring SSL support until support is added (either to Kubernetes or AWX itself).
- Docker on the host where AWX will be deployed. After installing Docker, the Docker service must be started (depending on your OS, you may have to add the local user that uses Docker to the
docker
group, refer to the documentation for details) - docker-py Python module.
- Docker Compose.
By default, the delivered installer/inventory file will deploy AWX to the local host. It is possible, however, to deploy to a remote host. The installer/install.yml playbook can be used to build images on the local host, and ship the built images to, and run deployment tasks on, a remote host. To do this, modify the installer/inventory file, by commenting out localhost
, and adding the remote host.
For example, suppose you wish to build images locally on your CI/CD host, and deploy them to a remote host named awx-server. To do this, add awx-server to the installer/inventory file, and comment out or remove localhost
, as demonstrated by the following:
# localhost ansible_connection=local
awx-server
[all:vars]
...
In the above example, image build tasks will be delegated to localhost
, which is typically where the clone of the AWX project exists. Built images will be archived, copied to remote host, and imported into the remote Docker image cache. Tasks to start the AWX containers will then execute on the remote host.
If you choose to use the official images then the remote host will be the one to pull those images.
Note
You may also want to set additional variables to control how Ansible connects to the host. For more information about this, view Behavioral Inventory Parameters.
As mentioned above, in Prerequisites, the prerequisites are required on the remote host.
When deploying to a remote host, the playbook does not execute tasks with the
become
option. For this reason, make sure the user that connects to the remote host has privileges to run thedocker
command. This typically means that non-privileged users need to be part of thedocker
group.
Before starting the build process, review the inventory file, and uncomment and provide values for the following variables found in the [all:vars]
section:
postgres_data_dir
If you're using the default PostgreSQL container (see PostgreSQL below), provide a path that can be mounted to the container, and where the database can be persisted.
host_port
Provide a port number that can be mapped from the Docker daemon host to the web server running inside the AWX container. Defaults to 80.
ssl_certificate
Optionally, provide the path to a file that contains a certificate and its private key.
docker_compose_dir
When using docker-compose, the docker-compose.yml
file will be created there (default /tmp/awxcompose
).
ca_trust_dir
If you're using a non trusted CA, provide a path where the untrusted Certs are stored on your Host.
If you wish to tag and push built images to a Docker registry, set the following variables in the inventory file:
docker_registry
IP address and port, or URL, for accessing a registry.
docker_registry_repository
Namespace to use when pushing and pulling images to and from the registry. Defaults to awx.
docker_registry_username
Username of the user that will push images to the registry. Defaults to developer.
docker_remove_local_images
Due to the way that the docker_image module behaves, images will not be pushed to a remote repository if they are present locally. Set this to delete local versions of the images that will be pushed to the remote. This will fail if containers are currently running from those images.
Note
These settings are ignored if using official images
http_proxy
IP address and port, or URL, for using an http_proxy.
https_proxy
IP address and port, or URL, for using an https_proxy.
no_proxy
Exclude IP address or URL from the proxy.
AWX requires access to a PostgreSQL database, and by default, one will be created and deployed in a container, and data will be persisted to a host volume. In this scenario, you must set the value of postgres_data_dir
to a path that can be mounted to the container. When the container is stopped, the database files will still exist in the specified path.
If you wish to use an external database, in the inventory file, set the value of pg_hostname
, and update pg_username
, pg_password
, pg_database
, and pg_port
with the connection information.
If you are not pushing images to a Docker registry, start the build by running the following:
# Set the working directory to installer
$ cd installer
# Run the Ansible playbook
$ ansible-playbook -i inventory install.yml
If you're pushing built images to a repository, then use the -e
option to pass the registry password as follows, replacing password with the password of the username assigned to docker_registry_username
(note that you will also need to remove dockerhub_base
and dockerhub_version
from the inventory file):
# Set the working directory to installer
$ cd installer
# Run the Ansible playbook
$ ansible-playbook -i inventory -e docker_registry_password=password install.yml
After the playbook run completes, Docker will report up to 5 running containers. If you chose to use an existing PostgresSQL database, then it will report 4. You can view the running containers using the docker ps
command, as follows:
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
e240ed8209cd awx_task:1.0.0.8 "/tini -- /bin/sh ..." 2 minutes ago Up About a minute 8052/tcp awx_task
1cfd02601690 awx_web:1.0.0.8 "/tini -- /bin/sh ..." 2 minutes ago Up About a minute 0.0.0.0:443->8052/tcp awx_web
55a552142bcd memcached:alpine "docker-entrypoint..." 2 minutes ago Up 2 minutes 11211/tcp memcached
84011c072aad rabbitmq:3 "docker-entrypoint..." 2 minutes ago Up 2 minutes 4369/tcp, 5671-5672/tcp, 25672/tcp rabbitmq
97e196120ab3 postgres:9.6 "docker-entrypoint..." 2 minutes ago Up 2 minutes 5432/tcp postgres
If you're deploying using Docker Compose, container names will be prefixed by the name of the folder where the docker-compose.yml file is created (by default, awx
).
Immediately after the containers start, the awx_task container will perform required setup tasks, including database migrations. These tasks need to complete before the web interface can be accessed. To monitor the progress, you can follow the container's STDOUT by running the following:
# Tail the the awx_task log
$ docker logs -f awx_task
You will see output similar to the following:
Using /etc/ansible/ansible.cfg as config file
127.0.0.1 | SUCCESS => {
"changed": false,
"db": "awx"
}
Operations to perform:
Synchronize unmigrated apps: solo, api, staticfiles, messages, channels, django_extensions, ui, rest_framework, polymorphic
Apply all migrations: sso, taggit, sessions, sites, kombu_transport_django, social_auth, contenttypes, auth, conf, main
Synchronizing apps without migrations:
Creating tables...
Running deferred SQL...
Installing custom SQL...
Running migrations:
Rendering model states... DONE
Applying contenttypes.0001_initial... OK
Applying contenttypes.0002_remove_content_type_name... OK
Applying auth.0001_initial... OK
Applying auth.0002_alter_permission_name_max_length... OK
Applying auth.0003_alter_user_email_max_length... OK
Applying auth.0004_alter_user_username_opts... OK
Applying auth.0005_alter_user_last_login_null... OK
Applying auth.0006_require_contenttypes_0002... OK
Applying taggit.0001_initial... OK
Applying taggit.0002_auto_20150616_2121... OK
Applying main.0001_initial... OK
...
Once migrations complete, you will see the following log output, indicating that migrations have completed:
Python 2.7.5 (default, Nov 6 2016, 00:28:07)
[GCC 4.8.5 20150623 (Red Hat 4.8.5-11)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> <User: admin>
>>> Default organization added.
Demo Credential, Inventory, and Job Template added.
Successfully registered instance awx
(changed: True)
Creating instance group tower
Added instance awx to tower
(changed: True)
...
The AWX web server is accessible on the deployment host, using the host_port value set in the inventory file. The default URL is http://localhost.
You will prompted with a login dialog. The default administrator username is admin
, and the password is password
.
After the installation, maintenance operations with docker-compose can be done by using the docker-compose.yml
file created at the location pointed by docker_compose_dir
.
Among the possible operations, you may:
- Stop AWX :
docker-compose stop
- Upgrade AWX :
docker-compose pull && docker-compose up --force-recreate
See the docker-compose documentation for details.