..note:: This section of the documentation is under construction. We are in the process of adding more examples about the Rackspace modules and how they work together. Once complete, there will also be examples for Rackspace Cloud in `ansible-examples <http://github.com/ansible/ansible-examples/>`_.
Most users will not be using virtualenv, but some users, particularly Python developers sometimes like to.
There are special considerations when Ansible is installed to a Python virtualenv, rather than the default of installing at a global scope. Ansible assumes, unless otherwise instructed, that the python binary will live at /usr/bin/python. This is done via the interpreter line in modules, however when instructed by setting the inventory variable 'ansible_python_interpreter', Ansible will use this specified path instead to find Python. This can be a cause of confusion as one may assume that modules running on 'localhost', or perhaps running via 'local_action', are using the virtualenv Python interpreter. By setting this line in the inventory, the modules will execute in the virtualenv interpreter and have available the virtualenv packages, specifically pyrax. If using virtualenv, you may wish to modify your localhost inventory definition to find this location as follows:
pyrax may be installed in the global Python package scope or in a virtual environment. There are no special considerations to keep in mind when installing pyrax.
The 'rax' module provides the ability to provision instances within Rackspace Cloud. Typically the provisioning task will be performed from your Ansible control server (in our example, localhost) against the Rackspace cloud API. This is done for several reasons:
- Avoiding installing the pyrax library on remote nodes
- No need to encrypt and distribute credentials to remote nodes
The rax module returns data about the nodes it creates, like IP addresses, hostnames, and login passwords. By registering the return value of the step, it is possible used this data to dynamically add the resulting hosts to inventory (temporarily, in memory). This facilitates performing configuration actions on the hosts in a follow-on task. In the following example, the servers that were successfully created using the above task are dynamically added to a group called "raxhosts", with each nodes hostname, IP address, and root password being added to the inventory.
Once your nodes are spun up, you'll probably want to talk to them again. The best way to handle his is to use the "rax" inventory plugin, which dynamically queries Rackspace Cloud and tells Ansible what nodes you have to manage. You might want to use this even if you are spinning up Ansible via other tools, including the Rackspace Cloud user interface. The inventory plugin can be used to group resources by metadata, region, OS, etc. Utilizing metadata is highly recommended in "rax" and can provide an easy way to sort between host groups and roles. If you don't want to use the ``rax.py`` dynamic inventory script, you could also still choose to manually manage your INI inventory file, though this is less recommended.
In Ansible it is quite possible to use multiple dynamic inventory plugins along with INI file data. Just put them in a common directory and be sure the scripts are chmod +x, and the INI-based ones are not.
To use the rackspace dynamic inventory script, copy ``rax.py`` into your inventory directory and make it executable. You can specify a credentails file for ``rax.py`` utilizing the ``RAX_CREDS_FILE`` environment variable.
..note:: Dynamic inventory scripts (like ``rax.py``) are saved in ``/usr/share/ansible/inventory`` if Ansible has been installed globally. If installed to a virtualenv, the inventory scripts are installed to ``$VIRTUALENV/share/inventory``.
..note:: Users of :doc:`tower` will note that dynamic inventory is natively supported by Tower, and all you have to do is associate a group with your Rackspace Cloud credentials, and it will easily synchronize without going through these steps::
As mentioned previously, you will often be running most of these modules outside of the host loop, and will need 'localhost' defined. The recommended way to do this, would be to create an ``inventory`` directory, and place both the ``rax.py`` script and a file containing ``localhost`` in it.
When utilizing a standard ini formatted inventory file (as opposed to the inventory plugin), it may still be adventageous to retrieve discoverable hostvar information from the Rackspace API.
Ansible is a powerful orchestration tool, and rax modules allow you the opportunity to orchestrate complex tasks, deployments, and configurations. The key here is to automate provisioning of infrastructure, like any other pice of software in an environment. Complex deployments might have previously required manaul manipulation of load balancers, or manual provisioning of servers. Utilizing the rax modules included with Ansible, one can make the deployment of additioanl nodes contingent on the current number of running nodes, or the configuration of a clustered applicaiton dependent on the number of nodes with common metadata. One could automate the following scenarios, for example:
* Servers that are removed from a Cloud Load Balancer one-by-one, updated, verified, and returned to the load balancer pool
* Expansion of an already-online environment, where nodes are provisioned, bootstrapped, configured, and software installed
* A procedure where app log files are uploaded to a central location, like Cloud Files, before a node is decommissioned
* Servers and load balancers that have DNS receords created and destroyed on creation and decomissioning, respectively
