Could not find the requested service "firewall"

[geerlingguy]

On macOS Sierra, running VirtualBox 5.1.8 and Vagrant 1.8.6, with Ansible 2.2.0, a fresh Drupal VM provision is bailing with:

TASK [geerlingguy.firewall : Ensure the firewall is enabled and will start on boot.] ***
fatal: [drupalvm]: FAILED! => {"changed": false, "failed": true, "msg": "Could not find the requested service \"'firewall'\": "}

RUNNING HANDLER [geerlingguy.firewall : restart firewall] **********************

PLAY RECAP *********************************************************************
drupalvm                   : ok=10   changed=6    unreachable=0    failed=1

This looks like it could be related to #829, but I'm doing a little more investigation. For now, the simplest fix is to stay on Ansible 2.1.x (e.g. pip install ansible==2.1) instead of installing Ansible 2.2.x... something's changed with Ansible's service module that must be causing this issue.

On Windows, the issue can be caused by non-UNIX line endings on the firewall service file (see #829 comments), but on Mac/Linux (and on Windows in general), it looks like Ansible 2.2's service module gets snagged, but only on Ubuntu 16.04, which leads me to believe something is missing for systemd.

Temporary Fixes

1. Switch back to Ansible 2.1.x (e.g. pip install ansible==2.1)
2. If you see this error, run vagrant reload --provision to restart the VM and continue provisioning)

[geerlingguy]

See also: geerlingguy/ansible-role-firewall#17

[geerlingguy]

Ha... then since the firewall is set up but not configured properly... it looks like SSH is being blocked, so fix #2 might not work that great. Or it could be an issue with my latest iteration of the geerlingguy/ubuntu1604 box, or with Vagrant 1.8.6. Or VMware 8.5. Grr.

[enapt]

Getting same error. :( Tried destorying box, copying from working config, etc, no luck.

[enapt]

After using vagrant reload --provision - I'm getting..

TASK [geerlingguy.postgresql : Ensure PostgreSQL packages are installed.] ******
fatal: [clbs]: FAILED! => {"failed": true, "msg": "'postgresql_packages' is undefined"}
    to retry, use: --limit @/vagrant/provisioning/playbook.retry

PLAY RECAP *********************************************************************
clbs                       : ok=106  changed=3    unreachable=0    failed=1   

[geerlingguy]

@enapthine - I opened geerlingguy/ansible-role-postgresql#7 for that issue—it seems I might need to adjust a few things in a few different roles for Ansible 2.2+ compatibility.

[geerlingguy]

I have a fix for this in the geerlingguy.firewall role. Will continue testing.

[geerlingguy]

See upstream issue: geerlingguy/ansible-role-firewall#22

[geerlingguy]

Heh... now that I fixed the firewall issue, I hit the same thing with the mailhog role:

TASK [geerlingguy.mailhog : Ensure mailhog is enabled and will start on boot.] *
fatal: [drupalvm]: FAILED! => {"changed": false, "failed": true, "msg": "Could not find the requested service \"'mailhog'\": "}

Will open another upstream role.

[geerlingguy]

And now that I fixed mailhog, I'm getting:

TASK [geerlingguy.java : Ensure Java is installed.] ****************************
fatal: [drupalvm]: FAILED! => {"failed": true, "msg": "'java_packages' is undefined"}

Ansible 2.2 is like a hydra for all my roles with custom init scripts and sneakily-defined variables :P

[geerlingguy]

Issue for Java role: geerlingguy/ansible-role-java#26