We’re using some good amount of Ansible playbooks in my company and we use TeamCity as our main CI service.
On TeamCity, our playbook output looks like this:
I don’t know about you, but I quite dislike the Ansible playbook default output. It’s quite packed and big and it’s hard to check where Ansible is currently working on.
TeamCity has a very nice feature called Build Script Interaction, where you format your stdout according to specific standards so that it can be interpreted by TeamCity.
This way you can open and close blocks, set variables, mark errors, etc.
Did I say „open and close blocks“? Yes I did and this would make the perfect feature for Ansible playbooks, right?
Luckily, Ansible can be extended by plugins and the Ansible output is handled by so called callback plugins. Callback plugins react to certain events (check out the class definition of CallbackBase for a list of available methods to override) and some of those events regard the start of plays or tasks.
So I extended the original default output plugin and override the specific methods for plays and tasks.
And with this plugin in place, the logs now look like this:
All nicely structured and TeamCity can even calculate the times of plays and tasks!