watcher/doc/source/contributor/plugin/planner-plugin.rst
lingyongxu 28b9766693 Update the documentation link for doc migration
This patch is proposed according to the Direction 10 of doc
migration(https://etherpad.openstack.org/p/doc-migration-tracking).

Change-Id: I4eb594115e350e28f9136f7003692a1ec0abfcf6
2017-09-18 09:19:33 +00:00

175 lines
6.0 KiB
ReStructuredText

..
Except where otherwise noted, this document is licensed under Creative
Commons Attribution 3.0 License. You can view the license at:
https://creativecommons.org/licenses/by/3.0/
.. _implement_planner_plugin:
===================
Build a new planner
===================
Watcher :ref:`Decision Engine <watcher_decision_engine_definition>` has an
external :ref:`planner <watcher_planner_definition>` plugin interface which
gives anyone the ability to integrate an external :ref:`planner
<watcher_planner_definition>` in order to extend the initial set of planners
Watcher provides.
This section gives some guidelines on how to implement and integrate custom
planners with Watcher.
.. _Decision Engine: watcher_decision_engine_definition
Creating a new plugin
=====================
First of all you have to extend the base :py:class:`~.BasePlanner` class which
defines an abstract method that you will have to implement. The
:py:meth:`~.BasePlanner.schedule` is the method being called by the Decision
Engine to schedule a given solution (:py:class:`~.BaseSolution`) into an
:ref:`action plan <action_plan_definition>` by ordering/sequencing an unordered
set of actions contained in the proposed solution (for more details, see
:ref:`definition of a solution <solution_definition>`).
Here is an example showing how you can write a planner plugin called
``DummyPlanner``:
.. code-block:: python
# Filepath = third-party/third_party/dummy.py
# Import path = third_party.dummy
from oslo_utils import uuidutils
from watcher.decision_engine.planner import base
class DummyPlanner(base.BasePlanner):
def _create_action_plan(self, context, audit_id):
action_plan_dict = {
'uuid': uuidutils.generate_uuid(),
'audit_id': audit_id,
'first_action_id': None,
'state': objects.action_plan.State.RECOMMENDED
}
new_action_plan = objects.ActionPlan(context, **action_plan_dict)
new_action_plan.create(context)
new_action_plan.save()
return new_action_plan
def schedule(self, context, audit_id, solution):
# Empty action plan
action_plan = self._create_action_plan(context, audit_id)
# todo: You need to create the workflow of actions here
# and attach it to the action plan
return action_plan
This implementation is the most basic one. So if you want to have more advanced
examples, have a look at the implementation of planners already provided by
Watcher like :py:class:`~.DefaultPlanner`. A list with all available planner
plugins can be found :ref:`here <watcher_planners>`.
Define configuration parameters
===============================
At this point, you have a fully functional planner. However, in more complex
implementation, you may want to define some configuration options so one can
tune the planner to its needs. To do so, you can implement the
:py:meth:`~.Loadable.get_config_opts` class method as followed:
.. code-block:: python
from oslo_config import cfg
class DummyPlanner(base.BasePlanner):
# [...]
def schedule(self, context, audit_uuid, solution):
assert self.config.test_opt == 0
# [...]
@classmethod
def get_config_opts(cls):
return super(
DummyPlanner, cls).get_config_opts() + [
cfg.StrOpt('test_opt', help="Demo Option.", default=0),
# Some more options ...
]
The configuration options defined within this class method will be included
within the global ``watcher.conf`` configuration file under a section named by
convention: ``{namespace}.{plugin_name}``. In our case, the ``watcher.conf``
configuration would have to be modified as followed:
.. code-block:: ini
[watcher_planners.dummy]
# Option used for testing.
test_opt = test_value
Then, the configuration options you define within this method will then be
injected in each instantiated object via the ``config`` parameter of the
:py:meth:`~.BasePlanner.__init__` method.
Abstract Plugin Class
=====================
Here below is the abstract ``BasePlanner`` class that every single planner
should implement:
.. autoclass:: watcher.decision_engine.planner.base.BasePlanner
:members:
:special-members: __init__
:noindex:
Register a new entry point
==========================
In order for the Watcher Decision Engine to load your new planner, the
latter must be registered as a new entry point under the
``watcher_planners`` entry point namespace of your ``setup.py`` file. If you
are using pbr_, this entry point should be placed in your ``setup.cfg`` file.
The name you give to your entry point has to be unique.
Here below is how you would proceed to register ``DummyPlanner`` using pbr_:
.. code-block:: ini
[entry_points]
watcher_planners =
dummy = third_party.dummy:DummyPlanner
.. _pbr: https://docs.openstack.org/pbr/latest
Using planner plugins
=====================
The :ref:`Watcher Decision Engine <watcher_decision_engine_definition>` service
will automatically discover any installed plugins when it is started. This
means that if Watcher is already running when you install your plugin, you will
have to restart the related Watcher services. If a Python package containing a
custom plugin is installed within the same environment as Watcher, Watcher will
automatically make that plugin available for use.
At this point, Watcher will use your new planner if you referenced it in the
``planner`` option under the ``[watcher_planner]`` section of your
``watcher.conf`` configuration file when you started it. For example, if you
want to use the ``dummy`` planner you just installed, you would have to
select it as followed:
.. code-block:: ini
[watcher_planner]
planner = dummy
As you may have noticed, only a single planner implementation can be activated
at a time, so make sure it is generic enough to support all your strategies
and actions.