Handling Failure during AMI creation

[Navid Paya]

Hi guys

I have a playbook that:

• Creates an EC2 instance
• Provisions the instance using a few roles
• Creates an AMI out of that instance
• Destroys the instance after the AMI is created successfully

Now the problem is if there's a failure somewhere after the instance creation, let's say npm is having issues which is quite normal or some ubuntu repository is having issues, and I run the playbook again, it will spin up a new instance. How can I handle this? Preferably a way to use the same instance if the failure happens before AMI creation part or at least automatically delete the instance if I hit an error before exiting the playbook successfully.

Here's my current playbook:

---

- name: Making preprations for the provisioning

  hosts: localhost

  connection: local

  gather_facts: no

  roles:

    - workstation

- name: Launch an Ubuntu 14.04 EC2 instance

  hosts: localhost

  connection: local

  gather_facts: no

  tasks:

  - name: Find the latest Ubuntu AMI

    ec2_ami_search: distro=ubuntu release=trusty region=ap-southeast-1 store=ebs-ssd virt=hvm

    register: ubuntu_image

  - name: Start the new EC2 instance

    ec2:

      image: "{{ ubuntu_image.ami }}"

      region: ap-southeast-1

      zone: ap-southeast-1b

      instance_type: t2.small

      vpc_subnet_id: subnet-xxxxxxxxx

      group_id: ['sg-xxxxxxxx', 'sg-xxxxxxxx']

      key_name: blahblah-deploy

      wait: yes

      wait_timeout: 500

      instance_tags:

        Name: basebox.exmaple.net

        Role: Blahblah Base AMI

    register: ec2

  - name: Add the new instance to host group

    add_host: hostname={{ item.private_ip }} groupname=launched

    with_items: ec2.instances

  - name: Wait for SSH to come up on the new instance

    wait_for: host={{ item.private_ip }} port=22 delay=60 timeout=320 state=started

    with_items: ec2.instances

- name: Copy the code to the new instance

  hosts: launched

  user: ubuntu

  sudo: true

  tasks:

  - synchronize: src=code/ dest=/tmp/code delete=yes archive=yes

- name: Provision the new instance to create the base AMI

  hosts: launched

  user: ubuntu

  sudo: true

  vars:

    app_user: blahblah

    app_root: /home/{{ app_user }}/apps

  roles:

    - nodesource.node

    - basebox

- name: Remove the code from the instance before creating the AMI

  hosts: launched

  user: ubuntu

  sudo: true

  tasks:

  - file: path=/tmp/code state=absent

- name: Create an AMI from the provisioned instance

  hosts: localhost

  connection: local

  gather_facts: no

  tasks:

  - ec2_ami: region=ap-southeast-1 instance_id={{ item.id }} wait=no name=basebox-2000000 wait=yes wait_timeout=300

    with_items:

      - "{{ ec2.instances }}"

- name: Delete the instance now that the AMI is created

  hosts: localhost

  connection: local

  gather_facts: no

  tasks:

  - ec2:

      region: ap-southeast-1

      instance_ids: "{{ ec2.instance_ids }}"

      state: 'absent'

I'm not looking for the complete answer. Just a clue how to do it.