Skip to content
This repository has been archived by the owner on Jun 10, 2024. It is now read-only.

Azure Module Guidelines

Zim Kalinowski edited this page Sep 13, 2018 · 4 revisions

General Coding Guidelines for Azure Modules

Note: This section is still under development

Input parameters

General Rules

  • options structure should be as flat as possible, preferably only 2 levels
  • option names should follow python_case
  • choice option names should follow python_case
  • resource name option should always be name not xxx_name (adding alias is allowed)
  • location option should be optional, if not specified, should be taken from the resource group
  • tags option should be implemented if resource supports tags
  • modules should support check_mode

Return values for main modules

Return values should include only minimal information:

  • resource id
  • any specific information allocated by the service that is important for the user

Idempotence

  • modules should be idempotent -- calling with the same options should not result in any change
  • there may be some exceptions, for instance passwords, when they can't be compared ith existing passwords. When password option is not specified, it should be ommited in idempotence check. Specifying password should cause update.

How to implement idempotence?

The most robust approach is as follows: (1) query current_resource_state structure/dictionary (2) construct new_resource_state from input parameters (3) compare new_resource_state with current_resource_state (4) update current_resource_state with new_resource_state (5) use updated current_resource_state to call create or create_or_update function

Please note that currently no module is entirely using this approach. Points (4) and (5) are important in order to preserve properties of resource that was not created by Ansible module and may have some properties that are not supported by current version of the module.

Testing

Tests that need to be performed on every module

  • create resource
  • create resource again with the same parameters, make sure no change was recorded
  • create resource again with different parameters, make sure proper changes were made and change was recorded
  • delete resource, make sure change was recorded
  • delete unexising resource, make sure no change was recorder

Testing Idempotence

  • every field that should be idempotent shall be tested using check_mode
  • single udpate test should be done with changes to all fields to check if they were updated correctly

Specific Guidelines for Azure Facts Modules

Input Parameters

Every facts module should at least contain following options. In most cases these options should be sufficient:

Option Description Optional
resource_group Resource group name YES
name Resource Name YES
tags Tags that should be matched YES

Facts module should allow users to:

  • query all the resources in a given subscription (if possible)
  • query all the resources in a specific resource group (if possible)
  • query specific resource instance
  • filter resources by tags (if possible)

Return value rules

Return values should be a composite of:

  • input of the main module
    • structure should be preserved as closely as possible
    • fields should be omitted only if impossible or difficult to retrieve
  • additional read only fields assigned by the service, for instance:
    • resource id field
    • assigned ip_address
    • any other fields that may be important to the user, like URLs etc.

Following rules should be followed:

  • return structure should have as short name as possible, for instance servers
  • return structure should always be a list of dictionaries, even when single resource is queried
  • every dictionary should be as flat as possible, no deep structures
  • field names should be in python_case rather than camelCase
  • field values should be exactly the same as equivalent input values of the main module (preferably both python_case)
  • return structure should always contain resource name
  • return structure should always contain resource_group name

Return values should be structured as in following example:

servers:
  - id: /subscriptions/xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx/resourceGroups/TestGroup/providers/Microsoft.DBforPostgreSQL/servers/postgreabdud1223
    resource_group: TestGroup
    name: postgreabdud1223
    location: eastus
    sku:
      name: GP_Gen4_2
      tier: GeneralPurpose
      capacity: 2
    version: "9.6"
    user_visible_state: Ready
    fully_qualified_domain_name: postgreabdud1223.postgres.database.azure.com

Documenting Return Values

Facts module should include full documentation of return value, as in following example:

RETURN = '''
servers:
    description: A list of dictionaries containing facts for PostgreSQL servers.
    returned: always
    type: complex
    contains:
        id:
            description:
                - Resource ID
            returned: always
            type: str
            sample: /subscriptions/xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx/resourceGroups/TestGroup/providers/Microsoft.DBforPostgreSQL/servers/postgreabdud1223
        resource_group:
            description:
                - Resource group name.
            returned: always
            type: str
            sample: TestGroup
        name:
            description:
                - Resource name.
            returned: always
            type: str
            sample: postgreabdud1223
        location:
            description:
                - The location the resource resides in.
            returned: always
            type: str
            sample: eastus
        sku:
            description:
                - The SKU of the server.
            returned: always
            type: complex
            contains:
                name:
                    description:
                        - The name of the SKU
                    returned: always
                    type: str
                    sample: GP_Gen4_2
                tier:
                    description:
                        - The tier of the particular SKU
                    returned: always
                    type: str
                    sample: GeneralPurpose
                capacity:
                    description:
                        - The scale capacity.
                    returned: always
                    type: int
                    sample: 2
        version:
            description:
                - Server version.
            returned: always
            type: str
            sample: "9.6"
        user_visible_state:
            description:
                - A state of a server that is visible to user.
            returned: always
            type: str
            sample: Ready
        fully_qualified_domain_name:
            description:
                - The fully qualified domain name of a server.
            returned: always
            type: str
            sample: postgreabdud1223.postgres.database.azure.com
'''

Samples

Samples should include all possible combinations of input parameters showing how to:

  • query all the resources of given type in an entire subscription
  • query all the resources of given type in the resource group
  • query specific resource
  • filter resources by tags

Testing

Following rules should apply to tests:

  • Do not create a separate test, add azure_rm_xx_facts tests to /test/integration/targets/azure_rm_xx module tests
  • Remember to add azure_rm_xx_facts to /test/integration/targets/azure_rm_xx/aliases
  • Do not create any additional instances of costly resources just to test facts - one is sufficient
  • Test all possible parameter combinations
  • Check if all expected return values are present in the structure
  • Test if querying unexisting single instance fails correctly

(ARchived) Working groups

Working groups are now in the Ansible forum

Ansible project:
Community, Contributor Experience, Docs, News, Outreach, RelEng, Testing

Cloud:
AWS, Azure, CloudStack, Container, DigitalOcean, Docker, hcloud, Kubernetes, Linode, OpenStack, oVirt, Virt, VMware

Networking:
ACI, AVI, F5, Meraki, Network, NXOS

Ansible Developer Tools:
Ansible-developer-tools

Software:
Crypto, Foreman, GDrive, GitLab, Grafana, IPA, JBoss, MongoDB, MySQL, PostgreSQL, RabbitMQ, Zabbix

System:
AIX, BSD, HP-UX, macOS, Remote Management, Solaris, Windows

Security:
Security-Automation, Lockdown

Tooling:
AWX, Galaxy, Molecule

Communities

Modules:
unarchive, xml

Plugins:
httpapi

Wiki

Roles, Communication, Reviewing, Checklist, TODO

Clone this wiki locally