playbook-error-handler04.yml

---

## Our goal is to create a playbook that perform pre-checks, changes, and post-checks.
## Upon a failure / error, the playbook will "ROLLBACK" any changes it has made thus far.
## This playbook can serve as a template for constructing 'intelligent' playbooks
## within your enterprise

- name: A playbook demonstrating Error Handling techniques
  hosts: planetexpress
  gather_facts: yes

  vars:
    # list of services to install (separated by commas and available in the apt repo)
    # apached = apache http server
    # vsftpd = sftp server
    apps_to_install: [apache2, vsftpd]

  tasks:
    ## our job is to install some software on ONLY Debian hosts within the network
    ## therefore our PRECHECK PHASE needs to involve a 'check' to ensure we ONLY are on
    ## Debian hosts
    - name: PRECHECK PHASE
      block:
           - name: Ensure the platforms we logged into are Debian
             fail:
               msg: "Ansible has detected this host is not part of the Debian family."
             when: ansible_os_family != "Debian"

      rescue:
           - name: PRECHECK PHASE - FAILED
             fail:
               msg: "PRECHECK PHASE - FAILED. Nothing to rollback. Exiting..."


    ## this is the block containing the work we actually want to do
    - name: MAINTENANCE PHASE
      block:
           - name: Install services(s) to our remote host(s)
             apt:
               state: present
               name: "{{ apps_to_install }}"
             become: yes

           - name: Turn up the service(s) installed on our remote host(s)
             service:
               name: "{{ item }}"
               state: started
             loop: "{{ apps_to_install }}"
             become: yes

      ## this rescue only runs if the BLOCK fails
      rescue:
           - name: Remove services(s) on our remote host(s)
             apt:
               state: absent
               name: "{{ apps_to_install }}"
             become: yes

           - name: MAINTENANCE PHASE - FAILED
             fail:
               msg: "MAINTENANCE PHASE - FAILED. Rollback complete. Exiting..."

    - name: POSTCHECK PHASE
      block:

           - name: populate service facts
             service_facts:

           - name: Print out the service facts
             debug:
               var: ansible_facts.services

           - name: Ensure all of the new services installed
             fail:
               msg: "Ansible has detected not all services have installed."
             when: ansible_facts.services.get(item).state != "running"
             loop: "{{ apps_to_install }}"

      ## this rescue only runs if the BLOCK fails
      rescue:
           - name: Remove services(s) on our remote host(s)
             apt:
               state: absent
               name: "{{ apps_to_install }}"
             become: yes

           - name: MAINTENANCE PHASE - FAILED
             fail:
               msg: "POSTCHECK PHASE - FAILED. Rollback complete. Exiting..."         

        ## in a real deployment you might consider an ALWAYS section to provide the results
        ## of how the playbook ran
        # always:
        #      - name: EXAMPLE - Email your team the results of the playbook
        #        mail:   # mail module is used to send SMTP (email)
        #           host:
        #           port:
        #           password:
        #           username:
        #           subject:
        #           body: send an email with results via the mail module