Kevin Carter 8e6dbd01c9 Convert existing roles into galaxy roles
This change implements the blueprint to convert all roles and plays into
a more generic setup, following upstream ansible best practices.

Items Changed:
* All tasks have tags.
* All roles use namespaced variables.
* All redundant tasks within a given play and role have been removed.
* All of the repetitive plays have been removed in-favor of a more
  simplistic approach. This change duplicates code within the roles but
  ensures that the roles only ever run within their own scope.
* All roles have been built using an ansible galaxy syntax.
* The `*requirement.txt` files have been reformatted follow upstream
  Openstack practices.
* Dynamically generated inventory is now more organized, this should assist
  anyone who may want or need to dive into the JSON blob that is created.
  In the inventory a properties field is used for items that customize containers
  within the inventory.
* The environment map has been modified to support additional host groups to
  enable the seperation of infrastructure pieces. While the old infra_hosts group
  will still work this change allows for groups to be divided up into seperate
  chunks; eg: deployment of a swift only stack.
* The LXC logic now exists within the plays.
* etc/openstack_deploy/user_variables.yml has all password/token
  variables extracted into the separate file
  etc/openstack_deploy/user_secrets.yml in order to allow seperate
  security settings on that file.

Items Excised:
* All of the roles have had the LXC logic removed from within them which
  should allow roles to be consumed outside of the `os-ansible-deployment`
  reference architecture.

Note:
* the directory rpc_deployment still exists and is presently pointed at plays
  containing a deprecation warning instructing the user to move to the standard
  playbooks directory.
* While all of the rackspace specific components and variables have been removed
  and or were refactored the repository still relies on an upstream mirror of
  Openstack built python files and container images. This upstream mirror is hosted
  at rackspace at "http://rpc-repo.rackspace.com" though this is
  not locked to and or tied to rackspace specific installations. This repository
  contains all of the needed code to create and/or clone your own mirror.

DocImpact
Co-Authored-By: Jesse Pretorius <jesse.pretorius@rackspace.co.uk>
Closes-Bug: #1403676
Implements: blueprint galaxy-roles
Change-Id: I03df3328b7655f0cc9e43ba83b02623d038d214e
2015-02-18 10:56:25 +00:00
..

Ansible OpenStack LXC Configuration

date

2013-09-05 09:51

tags

lxc, openstack, cloud, ansible

category

*nix

This directory contains the files needed to make the openstack_deployment process work. The inventory is generated from a user configuration file named openstack_user_config.yml. To load inventory you MUST copy the directory openstack_deploy to either $HOME/ or /etc/. With this folder in place, you will need to enter the folder and edit the file openstack_user_config.yml. The file will contain all of the IP addresses/hostnames that your infrastructure will exist on as well as a CIDR that your containers will have IP addresses assigned from. This allows for easy scaling as new nodes and or affinity for containers is all set within this file.

Please see the openstack_user_config.yml file in the provided /etc directory for more details on how that file is setup.

If you need some assistance defining the CIDR for a given ip address range check out http://www.ipaddressguide.com/cidr

Words on openstack_user_config.yml

While the openstack_user_config.yml file is noted fairly heavily with examples and information regarding the options, here's some more information on what the file consists of and how to use it.

Global options

The user configuration file has three globally available options. These options allow you to set the CIDR for all of your containers IP addresses, and a list of used IP addresses that you may not want the inventory system to collide with, global overrides which are added to inventory outside of "group_vars" and "var_files" files.


Global Options:
  • cidr:
  • used_ips:
  • global_overrides:

Here's the syntax for cidr.

cidr: <string>/<prefix>

To tell inventory not to attempt to consume IP addresses which may or may not exist within the defined cidr write all known IP addresses that are already consumed as a list in yaml format.

Heres the used_ips syntax

used_ips:
  - 10.0.0.250
  - 10.0.0.251
  - 10.0.0.252
  - 10.0.0.253

If you want to specify specific globally available options and do not want to place them in var_files or within the group_vars/all.yml file you can set them in a key = value par within the global_overrides hash.

Here's the global_overrides syntax

global_overrides:
  debug: True
  git_install_branch: master

Predefined host groups

The user configuration file has 4 defined groups which have mapping found within the openstack_environment.yml file.

The predefined groups are:
  • infra_hosts:
  • compute_hosts:
  • storage_hosts:
  • log_hosts:

Any host specified within these groups will have containers built within them automatically. The containers that will be build are all mapped out within the openstack_environment.json file.

When specifying hosts inside of any of the known groups the syntax is as follows:

infra_hosts:
  infra_host1:
    ip: 10.0.0.1

With this the top key is the host name and ip is used to set the known IP address of the host name. Even if you have the host names set within your environment using either the hosts file or a resolver you must specify the "ip".

If you want to use a host that is not in a predefined group and is used is some custom out of band Ansible play you can add a top level key for the host type with the host name and "ip" key. The syntax is the exact same as the predefined host groups.

Adding options to containers within targeted hosts

Within the host variables options can be added that will append to the host_vars of a given set of containers. This allows you to add "special" configuration to containers on a targeted host which may come in handy when scaling out or planning a deployment of services. To add these options to all containers within the host simply add container_vars under the host name and use key: value pairs for all of the desired options. All key: value pairs will be set as host_vars on all containers found under host name.

Here is an example of turning debug mode on all containers on infra1

infra_hosts:
  infra1:
    ip: 10.0.0.10
    container_vars:
      debug: True
  infra2:
    ...

In this example you can see that we are setting container_vars under the host name infra1 and that debug was set to True.

Limiting the container types:

When developing the inventory it may be useful to further limit the containers that will have access to the provided options. In this case you use the option limit_container_types followed by the type of container you with to limit the options to. When using the limit_container_types option the inventory script will perform a string match on the container name and if a match is found, even if it's a partial match, the options will be appended to the container.

Here is an example of adding cinder_backends to containers on a host named cinder1 under the storage_hosts group. The options will be limited to containers matching the type "cinder_volume".

storage_hosts:
  cinder1:
    ip: 10.0.0.10
    container_vars:
      cinder_backends:
        limit_container_types: cinder_volume
        lvm:
          volume_group: cinder-volumes
          driver: cinder.volume.drivers.lvm.LVMISCSIDriver
          backend_name: LVM_iSCSI
  cinder2:
    ...