Why is my idempotence action failing?

It is important to understand that Molecule does not do anything further than the default functionality of Ansible when determining if your tasks are idempotent or not. Molecule will simply run the converge action twice and check against Ansible’s standard output.

Therefore, if you are seeing idempotence failures, it is typically related to the underlying Ansible report and not Molecule.

If you are facing idempotence failures and intend to raise a bug on our issue tracker, please first manually run molecule converge twice and confirm that Ansible itself is reporting task idempotence (changed=0).

Why does Molecule make so many shell calls?

Ansible provides a Python API. However, it is not intended for direct consumption. We wanted to focus on making Molecule useful, so our efforts were spent consuming Ansible’s CLI.

Since we already consume Ansible’s CLI, we decided to call additional binaries through their respective CLI.


This decision may be reevaluated later.

Why does Molecule only support Ansible versions 2.2 and later?

  • Ansible 2.2 is the first good release in the Ansible 2 lineup.

  • The modules needed to support the drivers did not exist pre 2.2 or were not sufficient.

Why are playbooks used to provision instances?

Simplicity. Ansible already supports numerous cloud providers. Too much time was spent in Molecule v1, re-implementing a feature that already existed in the core Ansible modules.

Have you thought about using Ansible’s python API instead of playbooks?

This was evaluated early on. It was a toss up. It would provide simplicity in some situations and complexity in others. Developers know and understand playbooks. Decided against a more elegant and sexy solution.

Why are there multiple scenario directories and molecule.yml files?

Again, simplicity. Rather than defining an all encompassing config file opted to normalize. Molecule simply loops through each scenario applying the scenario’s molecule.yml.


This decision may be reevaluated later.

Are there similar tools to Molecule?

Can I run Molecule processes in parallel?

Please see Running Molecule processes in parallel mode for usage.

Can I specify random instance IDs in my molecule.yml?

This depends on the CI provider but the basic recipe is as follows.

Setup your molecule.yml to look like this:

  - name: "instance-${INSTANCE_UUID}"

Then in your CI provider environment, for example, Gitlab CI, setup:


Where CI_JOB_ID is the random variable that Gitlab provides.

Molecule will resolve the INSTANCE_UUID environment variable when creating and looking up the instance name. You can confirm all is in working order by running molecule list.

Can I test Ansible Collections with Molecule?

This is not currently officially supported. Also, collections remain in “tech preview” status. However, you can take a look at this blog post outlining a workable “DIY” solution as a stop gap for now.

Does Molecule support monorepos?

Yes, roles contained in a monorepo with other roles are automatically picked up and ANSIBLE_ROLES_PATH is set accordingly. See this page for more information.

How can I add development/testing-only dependencies?

Sometimes, it’s desirable to only run a dependency role when developing your role with molecule, but not impose a hard dependency on the role itself; for example when you rely on one of its side effects. This can be achieved by an approach like this in your role’s meta/main.yml:

  - role: <your-dependee-role>
    when: lookup('env', 'MOLECULE_FILE')