Additional Python Libraries

• libvirt-python: Libvirt Python SDK for the virt ansible module
• boto, boto3, botocore: AWS S3 Storage libraries

Ansible Integration Pack

This pack provides Ansible integration to perform remote operations on both local and remote machines. After pack installation all ansible executable files are available in pack virtualenv and ready to use.

Requirements

This pack installs Ansible from pip and therefore may require some OS-level packages to be in place.

Ansible

Ubuntu:

sudo apt install gcc libkrb5-dev

RHEL/CentOS:

sudo yum install gcc krb5-devel

Libvirt

sudo apt install libvirt-dev

Actions

• command - Run single Ad-Hoc command. It has all the regular parameters of ansible executable.
• command_local - Perform single ansible Ad-Hoc command (module) locally.
• playbook - Action to run Ansible Playbook (ansible-playbook executable).
• vault.encrypt - Encrypt ansible data files (playbooks, vars, roles, etc) with password (ansible-vault executable).
• vault.decrypt - Decrypt ansible data files (playbooks, vars, roles, etc) with password (ansible-vault executable).
• galaxy.install - Install role from Ansible Galaxy - hub of community developed roles (ansible-galaxy).
• galaxy.list - List installed from Ansible Galaxy roles (ansible-galaxy executable).
• galaxy.remove - Remove the role installed from Ansible Galaxy (ansible-galaxy executable).

Examples

See StackStorm with Ansible on Vagrant demo for more examples

ansible.command examples

# run ansible command with optional verbose parameter
st2 run ansible.command hosts=all args='hostname -i' verbose=vv

Action ansible.command_local is helper for the ansible.command with predefined parameters to run the command locally. So this is the same:

st2 run ansible.command_local args='echo $TERM'
st2 run ansible.command connection=local inventory_file='127.0.0.1,' hosts=all args='echo $TERM'

which is equivalent of ansible commands:

ansible all -c local -i '127.0.0.1,' -a 'echo $TERM'
ansible all --connection=local --inventory-file='127.0.0.1,' --args='echo $TERM'

ansible.playbook examples

# run some simple playbook
st2 run ansible.playbook playbook=/etc/ansible/playbooks/nginx.yml

# run playbook on last machine listed in inventory file
st2 run ansible.playbook playbook=/etc/ansible/playbooks/nginx.yml limit='all[-1]'

ansible.vault examples

# encrypt /tmp/nginx.yml playbook with password containing in vault.txt
st2 run ansible.vault.encrypt vault_password_file=vault.txt files=/tmp/nginx.yml

# decrypt /etc/ansible/nginx.yml and /etc/ansible/db.yml files
st2 run ansible.vault.decrypt cwd=/etc/ansible vault_password_file=vault.txt files='nginx.yml db.yml'

# decrypt all files in /etc/ansible/playbooks directory
st2 run ansible.vault.decrypt cwd=/etc/ansible vault_password_file=vault.txt files='playbooks/*'

ansible.galaxy examples

# download many roles
st2 run ansible.galaxy.install roles='bennojoy.mysql kosssi.composer'

# list rolex
st2 run ansible.galaxy.list roles_path=/etc/ansible/roles

Tips & Tricks

Using Ansible extra_vars in StackStorm Workflow

This is an example from a workflow that passes several different variables to the playbook as extra-vars:

sample_task:
  action: ansible.playbook
  input:
    playbook: /path/to/playbook.yml
    extra_vars:
      #
      # as key=value pairs
      - key1=value1
      - key2=value2
      #
      # variables from a yaml (or json) file
      - '@/path/to/file.yml'
      #
      # an arbitrarily complex dict of variables (passed as JSON to ansible)
      -
        key3: "{{ value3 }}"
        key4: [ value4a, value4b ]
        key5:
          - value5a
          - { value5bkey: value5bvalue }
        key6:
          key7: value7
          key8: value8

Structured output

# get structured JSON output from a playbook
st2 run ansible.playbook playbook=/etc/ansible/playbooks/nginx.yml env='{"ANSIBLE_STDOUT_CALLBACK":"json"}'

Using the JSON stdout_callback leads to JSON output which enables access to details of the result of the playbook in actions following the playbook execution, e.g. posting the results to Slack in an action-alias.

format: | 
    *Execution Overview*
    {% for host, result in execution.result.stdout.stats.iteritems() %}
        {{ host }}: ```{{ result }}```
    {% endfor %}

There is, however, a bug that breaks the JSON when the playbook execution fails (example output below). See this issue for more information. Manual handling of this case is necessary until the bug is fixed.

	to retry, use: --limit @/etc/ansible/playbooks/top.retry
{
    "plays": [
        {
            "play": {
                "id": "b5fe7b50-9d7d-4927-ac17-6886218bcabc", 
                "name": "some-host.com"
            }, 
            ...
}

Relative path to playbooks within StackStorm workflows

Current working directory (CWD) defaults to pack dir you're invoking Ansible pack actions from. That means if you're calling ansible.playbook from the custom.workflow, then you can use relative path to playbooks you'd ship with the custom pack (infra-as-code, yeah).

version: '2.0'
custom.workflow:
  description: A sample workflow that demonstrates how to use relative paths to playbooks shipped with pack.
  type: direct
  tasks:
    a:
      action: ansible.playbook
      input:
        # 'ansible_play.yml' is part of the 'custom' pack
        playbook: "ansible_play.yml"
        inventory_file: "localhost,"

This eliminates the need to specify absolute path to Ansible playbook file, located somewhere in /opt/stackstorm/packs/....

Windows Hosts

Connecting to windows is possibe as of version v0.5.2 of this pack. This is accomplished using ansible's builtin windows support.

Prior to executing a playbook on a Windows host, the host must be configured to accept WinRM connections. To accomplish this, execute the ansible setup PowerShell script on every Windows host you connect to. We recommend performing this on your Windows VM templates.

The following extra_vars must be passed in when executing a playbook on a Windows host:

• ansible_user : User to connect as (prefer [email protected] over domain\user)
• ansible_password : Password to use when connecting
• ansible_connection : Connection method to use (winrm for windows)
• ansible_port : Port to use for the connection (5986 for WinRM)
• ansible_winrm_transport : WinRM transport to use for the connection (suggested: ntlm or credssp, for more information consult the pywinrm documentation.
• ansible_winrm_server_cert_validation : Should the SSL cert be validated. (suggested: ignore)

Connecting via NTLM using a [email protected] style login:

st2 run ansible.playbook playbook=/etc/ansible/playbooks/windows_playbook.yaml inventory_file="winvm01.domain.tld," extra_vars='["[email protected]","ansible_password=xxx","ansible_port=5986","ansible_connection=winrm","ansible_winrm_server_cert_validation=ignore","ansible_winrm_transport=ntlm"]'

Connecting via CredSSP using a DOMAIN\user style login (note the extra \):

st2 run ansible.playbook playbook=/etc/ansible/playbooks/windows_playbook.yaml inventory_file="winvm01.domain.tld," extra_vars='["ansible_user=DOMAIN\\\\user","ansible_password=xxx","ansible_port=5986","ansible_connection=winrm","ansible_winrm_server_cert_validation=ignore","ansible_winrm_transport=credssp"]'