1
0
Fork 0
mirror of https://github.com/ansible-collections/community.general.git synced 2024-09-14 20:13:21 +02:00
community.general/docsite/rst/intro_networking.rst

229 lines
8.2 KiB
ReStructuredText
Raw Normal View History

Networking Support
==================
.. contents:: Topics
.. _working_with_networking_devices:
Working with Networking Devices
2016-05-24 19:46:33 +02:00
```````````````````````````````
Starting with Ansible version 2.1, you can now use the familiar Ansible models of playbook authoring and module development to manage heterogenous networking devices.
Ansible currently supports the following networking devices:
2016-05-06 21:59:07 +02:00
* Arista EOS
* Cisco NXOS
* Cisco IOS
* Cisco IOSXR
* Cumulus Linux
* Juniper JUNOS
* OpenSwitch
.. _networking_installation:
Network Automation Installation
2016-05-24 19:46:33 +02:00
```````````````````````````````
2016-05-24 19:46:33 +02:00
* Download the `latest Ansible network release <http://releases.ansible.com/ansible-network/latest/>`_.
* `Playbooks for testing <https://github.com/ansible/test-network-modules>`_ Ansible core network modules.
.. _networking_module_index:
Available Networking Modules
2016-05-24 19:46:33 +02:00
````````````````````````````
Most standard Ansible modules are designed to work with Linux/Unix or Windows machines and will not work with networking devices. Some modules (including "slurp", "raw", and "setup") are platform-agnostic and will work with networking devices.
To see what modules are available for networking devices, please browse the `"networking" section of the Ansible module index <http://docs.ansible.com/ansible/list_of_network_modules.html#>`_.
.. _understanding_provider_arguments:
Connecting to Networking Devices
2016-05-24 19:46:33 +02:00
````````````````````````````````
All core networking modules implement a *provider* argument, which is a collection of arguments used to define the characteristics of how to connect to the device. This section will assist in understanding how the provider argument is used.
Each core network module supports an underlying operating system and transport. The operating system is a one-to-one match with the module, and the transport maintains a one-to-many relationship to the operating system as appropriate. Some network operating systems only have a single transport option.
Each core network module supports some basic arguments for configuring the transport:
* host - defines the hostname or IP address of the remote host
* port - defines the port to connect to
* username - defines the username to use to authenticate the connection
* password - defines the password to use to authenticate the connection
* transport - defines the type of connection transport to build
* authorize - enables privilege escalation for devices that require it
* auth_pass - defines the password, if needed, for privilege escalation
Individual modules can set defaults for these arguments to common values that match device default configuration settings. For instance, the default value for transport is universally 'cli'. Some modules support other values such as EOS (eapi) and NXOS (nxapi), while some only support 'cli'. All arguments are fully documented for each module.
By allowing individual tasks to set the transport arguments independently, modules that use different transport mechanisms and authentication credentials can be combined as necessary.
One downside to this approach is that every task needs to include the required arguments. This is where the provider argument comes into play. The provider argument accepts keyword arguments and passes them through to the task to assign connection and authentication parameters.
The following two config modules are essentially identical (using nxos_config) as an example but it applies to all core networking modules::
---
nxos_config:
src: config.j2
host: "{{ inventory_hostname }}"
username: "{{ ansible_ssh_user }}"
password: "{{ ansible_ssh_pass }}"
transport: cli
---
vars:
cli:
host: "{{ inventory_hostname }}"
username: "{{ ansible_ssh_user }}"
password: "{{ ansible_ssh_pass }} "
transport: cli
nxos_config:
src: config.j2
provider: "{{ cli }}"
Given the above two examples that are equivalent, the arguments can also be used to establish precedence and defaults. Consider the following example::
---
vars:
cli:
host: "{{ inventory_hostname }}"
username: operator
password: secret
transport: cli
tasks:
- nxos_config:
src: config.j2
provider: "{{ cli }}"
username: admin
password: admin
In this example, the values of admin for username and admin for password will override the values of operator in cli['username'] and secret in cli['password'])
This is true for all values in the provider including transport. So you could have a singular task that is now supported over CLI or NXAPI (assuming the configuration is value). ::
---
vars:
cli:
host: "{{ inventory_hostname }}"
username: operator
password: secret
transport: cli
tasks:
- nxos_config:
src: config.j2
provider: "{{ cli }}"
transport: nxapi
If all values are provided via the provider argument, the rules for requirements are still honored for the module. For instance, take the following scenario::
---
vars:
conn:
password: cisco_pass
transport: cli
tasks:
- nxos_config:
src: config.j2
provider: "{{ conn }}"
Running the above task will cause an error to be generated with a message that required parameters are missing. ::
"msg": "missing required arguments: username,host"
Overall, this provides a very granular level of control over how credentials are used with modules. It provides the playbook designer maximum control for changing context during a playbook run as needed.
.. _networking_environment_variables:
Networking Environment Variables
2016-05-24 19:46:33 +02:00
````````````````````````````````
The following environment variables are available to Ansible networking modules:
username ANSIBLE_NET_USERNAME
password ANSIBLE_NET_PASSWORD
ssh_keyfile ANSIBLE_NET_SSH_KEYFILE
authorize ANSIBLE_NET_AUTHORIZE
auth_pass ANSIBLE_NET_AUTH_PASS
Variables are evaulated in the following order, listed from lowest to highest priority:
* Default
2016-05-06 21:59:07 +02:00
* Environment
* Provider
* Task arguments
.. _networking_module_conditionals:
Conditionals in Networking Modules
2016-05-24 19:46:33 +02:00
``````````````````````````````````
Ansible allows you to use conditionals to control the flow of your playbooks. Ansible networking command modules use the following unique conditional statements.
2016-05-06 21:59:07 +02:00
* eq - Equal
* neq - Not equal
* gt - Greater than
* ge - Greater than or equal
* lt - Less than
* le - Less than or equal
2016-05-06 21:59:07 +02:00
* contains - Object contains specified item
Conditional statements evalute the results from the commands that are
executed remotely on the device. Once the task executes the command
set, the waitfor argument can be used to evalute the results before
2016-05-25 16:29:02 +02:00
returning control to the Ansible playbook.
For example::
---
- name: wait for interface to be admin enabled
eos_command:
2016-05-25 22:37:21 +02:00
commands:
- show interface Ethernet4 | json
waitfor:
- "result[0].interfaces.Ethernet4.interfaceStatus eq connected"
In the above example task, the command :code:`show interface Ethernet4 | json`
is executed on the remote device and the results are evaluated. If
the path
:code:`(result[0].interfaces.Ethernet4.interfaceStatus)` is not equal to
"connected", then the command is retried. This process continues
2016-05-25 07:00:40 +02:00
until either the condition is satisfied or the number of retries has
expired (by default, this is 10 retries at 1 second intervals).
The commands module can also evaluate more than one set of command
results in an interface. For instance::
---
- name: wait for interfaces to be admin enabled
eos_command:
2016-05-25 22:37:21 +02:00
commands:
- show interface Ethernet4 | json
- show interface Ethernet5 | json
waitfor:
- "result[0].interfaces.Ethernet4.interfaceStatus eq connected"
- "result[1].interfaces.Ethernet4.interfaceStatus eq connected"
2016-05-25 07:00:40 +02:00
In the above example, two commands are executed on the
remote device, and the results are evaluated. By specifying the result
index value (0 or 1), the correct result output is checked against the
conditional.
The waitfor argument must always start with result and then the
2016-05-25 07:00:40 +02:00
command index in [], where 0 is the first command in the commands list,
1 is the second command, 2 is the third and so on.