Conditional continues to monitor result. Bug or Feature?

Archives Ansible Project
1

I have a playbook that checks to see if ansible can ssh to a VM. If it cannot, it assigns a role of deploy_vm_from_template to the host. During the process of deploying the VM the host gets booted, the network is configured and I verify that ansible can ssh to the host before continuing with other roles.

Everything worked fine, until I decided I needed to restart sssd after verifying that I could ssh to the host. What I found is that once, I could ssh to the host, every task in the role deploy_vm_from_template was skipped. It appears that the “when: result|failed” triggers when “result” is updated in my until loop in the role. I verified this by changing “result” in the playbook to “result_in_playbook” and the tasks after the until loop began running instead of being skipped.

My question, is this a bug or a feature? This seems like it could be a very powerful feature. It seems like result is a widely used variable. Should I be using more unique variables to avoid this situation?

From the playbook

pre_tasks:

  • name: “See if we can ssh to {{ inventory_hostname }} before attempting to build it”
    local_action: command ssh -o ConnectTimeout=10 servacct_ansible@{{ inventory_hostname }} hostname
    register: result
    ignore_errors: yes

roles:

  • { role: deploy_vm_from_template, when: result|failed }

#From the role

  • name: “Wait for {{ inventory_hostname }} to recognize servacct_ansible account”
    register: result
    until: result.rc == 0
    retries: 30
    delay: 10

  • name: Test ping module
    ping:

2

skipping a task still activates 'register' so subsequent tasks can
check 'result|skipped' and be conditioned on that fact.