Set up molecule testing

[p-j-smith]

• Add a centos7 molecule scenario (tests/molecule/centos7/molecule.yml) - it has two instances, one for a database and another for a web server
• Add an inventory for testing (tests/molecule/resources/inventory/hosts.yml)
• Add group vars for testing (tests/molecule/resources/inventory/group_vars/all.yml)
• Add a converge playbook for testing (tests/molecule/resources/converge.yml). This playbook uses tags to limit which roles are included. This means we can run molecule only on roles that have changed in a given pr.
• Add a workflow to run molecule on mirsg.infrastructure.provision when the provision role has changed
• Add a requirements file so we can install this collection

Edit: @drmatthews after talking offline, we'll use a single instance for running molecule tests rather than two (database + web server)

[p-j-smith]

this limits molecule to run only mirsg.infrastructure.provision

[p-j-smith]

if we don't define MOLECULE_RUN_TAGS, all converge steps are performed

[p-j-smith]

we will need to add the other roles here, and probably target specific groups for some of them

[drmatthews]

Hi @p-j-smith - I don't think we need collection level testing here. We do it in the xnat collection because we have a playbook we want to test (i.e. we want to know that the roles work in combination). Here each of the roles has its own test and in CI we want those tests to run when changes are made to the role itself (as you have done already with the path filter in the workflow). Does that makes sense?

[drmatthews]

👍