- By: Toni Schmidbauer ( Lastmod: 2021-08-29 )
Every wondered why Ansible Tower does not start downloading required collections when you synchronize a project? Here are the stumbling blocks we discovered so far:
When downloading collections Ansible Tower searches for a file
requirements.yml in the collections directory.
Be careful with the file extension:
requirements.yml has to end with
.yml and not
Within Ansible Tower there is a setting called
Jobs. This has to be set to true, which
is also the default.
Last but not least an Ansible Galaxy credential needs to be defined
for the organization where the project is defined. With the default
installation of Ansible Tower, when the sample playbooks are installed
there is a credential called
Ansible Galaxy defined. You need to assign
this credential to the organization.
If you skip installing the sample playbooks, no
credential will be defined for you and you have to create it manually.
Ansible Tower uses a Python virtual environment for running
Ansible. The default environment is installed in
/var/lib/awx/venv/awx. You can also create custom environments, see
Using virtualenv with Ansible Tower.
In the default setup the following files define how collections are downloaded:
task.py defines various internal tasks Tower has to run on various
occasions. For example in line number 1930 (Ansible Tower 3.8.3) the
RunProjectUpdate gets defined. This is the task Tower
has to run whenever a project update is required.
In our case the function
build_extra_vars_file (line 2083 with
Ansible Tower 3.8.3) defines the variable
only if the organization has a galaxy credential defined (line 2099
Ansible Tower 3.8.3).
Line 2120 (Ansible Tower 3.8.3) finally defines the Ansible extra
collections_enabled depending on
task.py defines the extra variable
above). Finally the playbook
project_update.yml consumes this extra
variable and only downloads collections if
true, see the block string at line 192 (Ansible Tower 3.8.3)
So long and thanks for all the fish!