Molecule: vagrant driver writes vagrant-console-log to installation dir instead of $TMPDIR

[pescobar]

Issue Type

• Bug report

Molecule and Ansible details

ansible 2.7.1
molecule, version 2.19.0
python-vagrant==0.5.15

Molecule installation method (one of):

• pip

Ansible installation method (one of):

• pip

Desired Behavior

molecule should write all the logs to $TMPDIR. If molecule is installed in a folder where you don't have write permissons it crashes when booting a vagrant instance.

how to reproduce the problem

$> echo $TMPDIR
/tmp/escobar

$> molecule init role --driver-name vagrant --role-name testing-molecule-vagrant

$> rm  testing-molecule-vagrant/tasks/main.yml

$> cat <<EOF >> testing-molecule-vagrant/tasks/main.yml
---
# tasks file for testing-molecule-vagrant

- name: just a test
  command: uname -r
  changed_when: false
EOF

$> cd testing-molecule-vagrant

$> molecule --debug test

During TASK [Create molecule instance(s)] you get error: "msg": "ERROR: See log file '/tmp/escobar/molecule/testing-molecule-vagrant/default/vagrant-instance.err'"

and in the log file:

There was an error while executing `VBoxManage`, a CLI used by Vagrant
for controlling VirtualBox. The command and stderr is shown below.

Command: ["startvm", "384da504-e582-41a5-8f4f-08f478b62768", "--type", "headless"]

Stderr: VBoxManage: error: RawFile#0 failed to create the raw output file /data/virtualenvs/ansible-2.7.1-molecule-2.19.0/lib/python2.7/site-packages/molecule/provisioner/ansible/playbooks/vagrant/ubuntu-xenial-16.04-cloudimg-console.log (VERR_ACCESS_DENIED)
VBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component ConsoleWrap, interface IConsole
### 2018-11-05 22:06:45 ###

in this machine ansible and molecule are installed in a virtualenv located in /data/virtualenvs/ansible-2.7.1-molecule-2.19.0/ that is not owned by my user

[pescobar]

I found I can workaround the problem if I change default/molecule.yml from the default value:

platforms:
  - name: instance
    box: ubuntu/xenial64

to

platforms:
  - name: instance
    box: bento/ubuntu-16.04

I thought that the log file is a vagrant option and it's independent of the used box but it seems it's not like that?

[jonashackt]

I had the exact same problem - using another box worked also for me.

BUT I wanted to use my "standard" ubuntu/bionic64 or ubuntu/xenial64, so I dived a bit deeper. And there´s a reported issue with this fix here: https://github.com/joelhandwell/ubuntu_vagrant_boxes/issues/1#issuecomment-292370353, where you need to add v.customize [ "modifyvm", :id, "--uartmode1", "disconnected" ] to your Vagrantfile. Since we´re using Molecule which generates the Vagrantfile for us, we need to use platform.provider_raw_config_args to do the trick:

platforms:
  - name: ubuntu-docker
    box: ubuntu/bionic64
    memory: 512
    cpus: 1
    provider_raw_config_args:
    - "customize [ 'modifyvm', :id, '--uartmode1', 'disconnected' ]"

Here´s my example repo for checking full source: https://github.com/jonashackt/molecule-ansible-docker-vagrant

I consider this fix something that could be useful to have in the default Molecule Vagrant module because it isn´t quite beautiful to have to always set this up to get things running - especially for Molecule beginners.