How to use add_hosts with ec2 inventory?

[Mark]

Hello!

I'm calling a playbook with:

ansible-playbook provision.yml -i ./ec2.py --extra-vars "role=webserver env=dev01 hosts=type_t1_micro"

And in my playbook I want to add the hosts in the type_t1_micro group to the 'webservers' group, e.g.:

  tasks:
  - name: Adding new instance to host group
    local_action: add_host hostname=${item} groupname=webservers
    with_items: $hosts

This doesn't work because it tries to use 'type_t1_micro' as a host name - it doesn't add the actual hosts in the group.

How can I do this?

Thanks

[Michael DeHaan]

public service announcement.

So in Ansible 1.2 and later there is no need to do "$hosts"

with_items: hosts

you should also do hostname={{ item }}.

I'm not sure where you set "hosts" though, perhaps you can show the rest of the playbook?

--
You received this message because you are subscribed to the Google Groups "Ansible Project" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ansible-proje...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

--

Michael DeHaan <mic...@ansibleworks.com>
CTO, AnsibleWorks, Inc.
http://www.ansibleworks.com/

[Mark]

I've been having issues with the '{{ var }}' format and yaml compatibility so I've had to mix-and-match ${ var } syntax. Perhaps I just always need to use curlies quoted like "{{ var }}"?

I might be misunderstanding with setting $hosts, so here's my complete playbook (launch-ec2.yml):

# Launches an EC2 instance and provisions it as the specified role
#
# A particular role can be provisioned with, e.g.:
# ansible-playbook launch-ec2.yml -i localhost --extra-vars "role=webserver env=dev01"
- name: Launching an EC2 instance then provisioning it as a {{ role }}
  vars_files:
    - vars/all.yml
    - vars/$env.yml           # include the env-specific variables (e.g. dev01.yml)
  hosts: localhost
  gather_facts: False
  connection: local
  tasks:
    - name: Launching instance
      local_action:
        module: ec2
        keypair: ${aws.keypair}
        group: ${aws.security_group}
        instance_type: ${aws.instance_type}
        image: ${aws.image}
        wait: true
        region: ${aws.region}
        instance_tags: '{"Name": "{{ role }}", "env": "{{ env }}"}'
        user_data: "{{ lookup('file', 'files/cloudinit') }}"
        ec2_access_key: ${aws.access_key_id}
        ec2_secret_key: ${aws.secret_access_key}
      register: ec2

    - name: Adding new instance to host group

      local_action: add_host hostname=${item.public_dns_name} groupname=webservers
      with_items: ec2.instances
    - name: Waiting for SSH to come up
      local_action: wait_for host=${item.public_dns_name} port=22 delay=10 timeout=120 state=started
      with_items: ec2.instances

- include: provision.yml hosts=webservers env=env

And provision.yml is:

# Provisions a running instance as the role supplied by extra-vars.
#
# A running EC2 instance can be provisioned directly with, e.g.:
# ansible-playbook provision.yml -i ./ec2.py --extra-vars "role=webserver env=dev01 user=ubuntu hosts=tag_Name_webserver"

- name: Configuring instance(s) as a {{ role }}
  vars_prompt:
    role: "Enter role to configure"
    env: "Enter environment name"
  vars_files:       # need this here too in case we're using vagrant
    - vars/all.yml
    - vars/$env.yml           # include the env-specific variables (e.g. dev01.yml)
  hosts: $hosts
  sudo: True
  gather_facts: True
  roles:  # provision the instance as the appropriate role, passed as a parameter to this play
    - base
    - $role

   # this is where I want to dynamically add the $hosts to 'webservers', ready for the deploy-site.yml playbook

- include: deploy-site.yml

And deploy-site.yml is:

- name: Deploy the web app to the web servers
  vars_files:       # need this here too in case we're using vagrant
    - vars/all.yml
    - vars/$env.yml           # include the env-specific variables (e.g. dev01.yml)
    - roles/webserver/vars/main.yml
  hosts: webservers             # only target webservers
  user: $user
  sudo: True
  sudo_user: web
  gather_facts: True
  tasks:
  - include: roles/webserver/tasks/deploy.yml

I've set these up so that I can either call launch-ec2.yml directly, or I can use vagrant just with provision.yml. Here's my Vagrantfile:

Vagrant.configure("2") do |config|

    config.vm.provision :ansible do |ansible|
        ansible.playbook = "ansible/provision.yml"
        # uncomment the following line to only run the deployment
#        ansible.playbook = "ansible/deploy-site.yml"
        ansible.inventory_file = "ansible/vagrant-hosts"
        ansible.verbose = true
        ansible.extra_vars = {
            hosts: "vagrant",
            role: "webserver",
            user: "vagrant",
            env: "vagrant"
        }
    end
    ...

Also, if I want I can reprovision either a vagrant or an EC2 instance by just calling provision.yml or deploy-site.yml.

Perhaps the issue is that I'm only targetting webservers in deploy-site.yml (because that's all I want to deploy to for this playbook)? I tried using 'all' in combination with a --limit parameter to ansible-playbook but then ansible started trying to configure 'localhost'.

So that's where I thought if I could pass a host-group from the ec2.py inventory script, but then add all hosts in it to 'webservers', that might work, but I don't seem able to...