4c150b39c1
The default value should be None, for "unset" not False. |
||
|---|---|---|
| .. | ||
| README.md | ||
| __init__.py | ||
| _os_server_actions.py | ||
| os_auth.py | ||
| os_client_config.py | ||
| os_coe_cluster.py | ||
| os_coe_cluster_template.py | ||
| os_flavor_facts.py | ||
| os_floating_ip.py | ||
| os_group.py | ||
| os_image.py | ||
| os_image_facts.py | ||
| os_ironic.py | ||
| os_ironic_inspect.py | ||
| os_ironic_node.py | ||
| os_keypair.py | ||
| os_keystone_domain.py | ||
| os_keystone_domain_facts.py | ||
| os_keystone_endpoint.py | ||
| os_keystone_role.py | ||
| os_keystone_service.py | ||
| os_listener.py | ||
| os_loadbalancer.py | ||
| os_member.py | ||
| os_network.py | ||
| os_networks_facts.py | ||
| os_nova_flavor.py | ||
| os_nova_host_aggregate.py | ||
| os_object.py | ||
| os_pool.py | ||
| os_port.py | ||
| os_port_facts.py | ||
| os_project.py | ||
| os_project_access.py | ||
| os_project_facts.py | ||
| os_quota.py | ||
| os_recordset.py | ||
| os_router.py | ||
| os_security_group.py | ||
| os_security_group_rule.py | ||
| os_server.py | ||
| os_server_action.py | ||
| os_server_facts.py | ||
| os_server_group.py | ||
| os_server_metadata.py | ||
| os_server_volume.py | ||
| os_stack.py | ||
| os_subnet.py | ||
| os_subnets_facts.py | ||
| os_user.py | ||
| os_user_facts.py | ||
| os_user_group.py | ||
| os_user_role.py | ||
| os_volume.py | ||
| os_volume_snapshot.py | ||
| os_zone.py | ||
README.md
OpenStack Ansible Modules
These are a set of modules for interacting with OpenStack as either an admin or an end user. If the module does not begin with os_, it's either deprecated or soon to be. This document serves as developer coding guidelines for modules intended to be here.
Naming
- All modules should start with os_
- If the module is one that a cloud consumer would expect to use, it should be named after the logical resource it manages. Thus, os_server not os_nova. The reasoning for this is that there are more than one resource that are managed by more than one service and which one manages it is a deployment detail. A good example of this are floating IPs, which can come from either Nova or Neutron, but which one they come from is immaterial to an end user.
- If the module is one that a cloud admin would expect to use, it should be be named with the service and the resource, such as os_keystone_domain.
- If the module is one that a cloud admin and a cloud consumer could both use, the cloud consumer rules apply.
Interface
- If the resource being managed has an id, it should be returned.
- If the resource being managed has an associated object more complex than an id, it should also be returned.
Interoperability
- It should be assumed that the cloud consumer does not know a bazillion details about the deployment choices their cloud provider made, and a best effort should be made to present one sane interface to the ansible user regardless of deployer insanity.
- All modules should work appropriately against all existing known public OpenStack clouds.
- It should be assumed that a user may have more than one cloud account that they wish to combine as part of a single ansible managed infrastructure.
Libraries
- All modules should use openstack_full_argument_spec to pick up the standard input such as auth and ssl support.
- All modules should extends_documentation_fragment: openstack to go along with openstack_full_argument_spec.
- All complex cloud interaction or interoperability code should be housed in the openstacksdk library.
- All OpenStack API interactions should happen via the openstacksdk and not via OpenStack Client libraries. The OpenStack Client libraries do no have end users as a primary audience, they are for intra-server communication.
Testing
- Integration testing is currently done in OpenStack's CI system in https://git.openstack.org/cgit/openstack/openstacksdk/tree/openstack/tests/ansible
- Testing in openstacksdk produces an obvious chicken-and-egg scenario. Work is under way to trigger from and report on PRs directly.