Molecule: Question: Regarding group_vars path linking and overriding

[r0ckyte]

Molecule and Ansible details

ansible --version
ansible 2.4.0.0

molecule --version
molecule, version 2.3.1

• Molecule installation method: pip
• Ansible installation method: pip

our git repo is configured as below

kerberos.yml
group_vars/all.yml
dev/group_vars/all.yml
qa/group_vars/all.yml
prod/group_vars/all.yml
roles/
hive
hbase
ruby
molecule/default/molecule.yml
webserver.yml
mysql.yml

i'm looking if its possible to override variables like first look in dev/group_vars , then qa/group_vars ..so on.

currently i'm declaring all variables required in molecule.yml and am able to run it good but i want to avoid this type of declaration.

Please suggest if any improvements can be incorporated

[retr0h]

Hi @r0ckyte this is not possible. Ansible loads group vars based on the group the system is in, and Molecule will only symlink out to a single group_vars directory.

You can try moving all of your groups under a single directory, such as:

group_vars/dev/all.yml
group_vars/qa/all.yml
group_vars/prod/all.yml

Then placing your systems into groups, but I'm not so sure how that would work. Also, you really want to test these things in isolation a bit more. You don't want items from dev bleeding over into your qa tests.

platforms:
  - name: etcd-01
    image: solita/ubuntu-systemd:latest
    privileged: True
    command: /sbin/init
    groups:
      - dev
      - qa
      - prod

I suggest a scenario per scenario you wish to test.

[r0ckyte]

@retr0h - Thanks for the suggestion.
lastly, few of my roles has include_vars as first statement and molecule is searching specified vars file at .molecule/group_vars not at path specified. can i override include_vars via molecule.yml ?

This is a great framework to test, i'm learning slowly and enjoying it. Thanks a ton !!

[retr0h]

@retr0h - Thanks for the suggestion.
lastly, few of my roles has include_vars as first statement and molecule is searching specified vars file at .molecule/group_vars not at path specified. can i override include_vars via molecule.yml ?

That I am not sure of. I believe Ansible is using that directory b/c Ansible by default uses the directory specified by -i. Molecule has to maintain host inventory so it uses -i molecule/$scenario_name/.molecule/ansible_inventory.yml. I'm not entirely sure how one would change that behavior.

[retr0h]

Looks like you could get fancy with include_vars and pwd.

tasks:
  - include_vars:
      file: "{{ lookup('env','PWD') }}/foo.yml"

The above will load foo.yml from the current directory. You can modify to suit your needs. I also suggest looking into other ansible built in vars such as playbook_dir.

[retr0h]

@r0ckyte thank you for the issue. Closing out.