using custom vars in group_by

[O. T. Suarez]

Hi,

I want to be able to execute tasks on a subgroup of servers. It can be done via groups ([dbservers_cannary]) in the inventory file, but after checking out group_by funcionality, I thought of just adding a variable to the host in order to identify its pertenence to the subgroup.

This is for the cannary release before full deploy.

The code below doesn't work as originally intented. Any idea what I am missing?

Regards,

Osvaldo

$ cat inventory

[dbservers]

db1 ansible_connection=local deploy=release

db2 ansible_connection=local deploy=cannary

db3 ansible_connection=local deploy=release

[wwwservers]

web1 ansible_connection=local deploy=cannary

web2 ansible_connection=local deploy=release

web3 ansible_connection=local deploy=release

$ cat vars.yml

---

- hosts: all

  gather_facts: false

  tasks:

  - group_by: key=deploy

- name: ping cannary servers

  hosts: cannary

  gather_facts: false

  tasks:

  - name: ping

    action: ping

$ ansible-playbook -i inventory vars.yml

PLAY [all] ******************************************************************** 

TASK: [group_by key=deploy] *************************************************** 

changed: [web1]

PLAY [ping cannary servers] *************************************************** 

skipping: no hosts matched

PLAY RECAP ******************************************************************** 

db1                        : ok=1    changed=1    unreachable=0    failed=0   

...

$