Correct Doc location and update for middleware only
Correct the location of the documents and only publish middleware documentation.
0
docs/.gitignore → doc/.gitignore
vendored
@ -62,9 +62,9 @@ qthelp:
|
||||
@echo
|
||||
@echo "Build finished; now you can run "qcollectiongenerator" with the" \
|
||||
".qhcp project file in $(BUILDDIR)/qthelp, like this:"
|
||||
@echo "# qcollectiongenerator $(BUILDDIR)/qthelp/python-keystoneclient.qhcp"
|
||||
@echo "# qcollectiongenerator $(BUILDDIR)/qthelp/keystonemiddleware.qhcp"
|
||||
@echo "To view the help file:"
|
||||
@echo "# assistant -collectionFile $(BUILDDIR)/qthelp/python-keystoneclient.qhc"
|
||||
@echo "# assistant -collectionFile $(BUILDDIR)/qthelp/keystonemiddleware.qhc"
|
||||
|
||||
latex:
|
||||
$(SPHINXBUILD) -b latex $(ALLSPHINXOPTS) $(BUILDDIR)/latex
|
@ -35,10 +35,10 @@ def run_apidoc(app):
|
||||
run_already = True
|
||||
|
||||
package_dir = path.abspath(path.join(app.srcdir, '..', '..',
|
||||
'keystoneclient'))
|
||||
'keystonemiddleware'))
|
||||
source_dir = path.join(app.srcdir, 'api')
|
||||
apidoc.main(['apidoc', package_dir, '-f',
|
||||
'-H', 'keystoneclient Modules',
|
||||
'-H', 'keystonemiddleware Modules',
|
||||
'-o', source_dir])
|
||||
|
||||
|
@ -1,6 +1,6 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
#
|
||||
# python-keystoneclient documentation build configuration file, created by
|
||||
# keystonemiddleware documentation build configuration file, created by
|
||||
# sphinx-quickstart on Sun Dec 6 14:19:25 2009.
|
||||
#
|
||||
# This file is execfile()d with the current directory set to its containing
|
||||
@ -64,13 +64,13 @@ source_suffix = '.rst'
|
||||
master_doc = 'index'
|
||||
|
||||
# General information about the project.
|
||||
project = 'python-keystoneclient'
|
||||
project = 'keystonemiddleware'
|
||||
copyright = 'OpenStack Contributors'
|
||||
|
||||
# The version info for the project you're documenting, acts as replacement for
|
||||
# |version| and |release|, also used in various other places throughout the
|
||||
# built documents.
|
||||
version_info = pbr.version.VersionInfo('python-keystoneclient')
|
||||
version_info = pbr.version.VersionInfo('keystonemiddleware')
|
||||
# The short X.Y version.
|
||||
version = version_info.version_string()
|
||||
# The full version, including alpha/beta/rc tags.
|
||||
@ -117,10 +117,7 @@ pygments_style = 'sphinx'
|
||||
# Grouping the document tree for man pages.
|
||||
# List of tuples 'sourcefile', 'target', 'title', 'Authors name', 'manual'
|
||||
|
||||
man_pages = [
|
||||
('man/keystone', 'keystone', 'Client for OpenStack Identity API',
|
||||
['OpenStack Contributors'], 1),
|
||||
]
|
||||
man_pages = []
|
||||
|
||||
# -- Options for HTML output --------------------------------------------------
|
||||
|
||||
@ -195,7 +192,7 @@ html_last_updated_fmt = os.popen(git_cmd).read()
|
||||
#html_file_suffix = ''
|
||||
|
||||
# Output file base name for HTML help builder.
|
||||
htmlhelp_basename = 'python-keystoneclientdoc'
|
||||
htmlhelp_basename = 'keystonemiddlewaredoc'
|
||||
|
||||
|
||||
# -- Options for LaTeX output -------------------------------------------------
|
||||
@ -210,8 +207,8 @@ htmlhelp_basename = 'python-keystoneclientdoc'
|
||||
# (source start file, target name, title, author, documentclass [howto/manual])
|
||||
# .
|
||||
latex_documents = [
|
||||
('index', 'python-keystoneclient.tex',
|
||||
'python-keystoneclient Documentation',
|
||||
('index', 'keystonmiddleware.tex',
|
||||
'keystonemiddleware Documentation',
|
||||
'Nebula Inc, based on work by Rackspace and Jacob Kaplan-Moss',
|
||||
'manual'),
|
||||
]
|
Before Width: | Height: | Size: 2.9 KiB After Width: | Height: | Size: 2.9 KiB |
Before Width: | Height: | Size: 3.5 KiB After Width: | Height: | Size: 3.5 KiB |
34
doc/source/index.rst
Normal file
@ -0,0 +1,34 @@
|
||||
Python Middleware for OpenStack Identity API (Keystone)
|
||||
=======================================================
|
||||
|
||||
This is the middleware provided for integrating with the OpenStack
|
||||
Identity API and handling authorization enforcement based upon the
|
||||
data within the OpenStack Identity tokens.
|
||||
|
||||
Contents:
|
||||
|
||||
.. toctree::
|
||||
:maxdepth: 1
|
||||
|
||||
middlewarearchitecture
|
||||
|
||||
Contributing
|
||||
============
|
||||
|
||||
Code is hosted `on GitHub`_. Submit bugs to the Keystone project on
|
||||
`Launchpad`_. Submit code to the ``openstack/keystonemiddleware`` project
|
||||
using `Gerrit`_.
|
||||
|
||||
.. _on GitHub: https://github.com/openstack/keystonemiddleware
|
||||
.. _Launchpad: https://launchpad.net/keystonemiddleware
|
||||
.. _Gerrit: http://wiki.openstack.org/GerritWorkflow
|
||||
|
||||
Run tests with ``python setup.py test``.
|
||||
|
||||
Indices and tables
|
||||
==================
|
||||
|
||||
* :ref:`genindex`
|
||||
* :ref:`modindex`
|
||||
* :ref:`search`
|
||||
|
@ -32,7 +32,7 @@ authentication middleware which acts as the internal API mechanism for
|
||||
OpenStack projects based on the WSGI standard.
|
||||
|
||||
This documentation describes the implementation in
|
||||
:class:`keystoneclient.middleware.auth_token`
|
||||
:class:`keystonemiddleware.auth_token`
|
||||
|
||||
Specification Overview
|
||||
======================
|
||||
@ -120,7 +120,7 @@ a WSGI component. Example for the auth_token middleware::
|
||||
pipeline = authtoken myService
|
||||
|
||||
[filter:authtoken]
|
||||
paste.filter_factory = keystoneclient.middleware.auth_token:filter_factory
|
||||
paste.filter_factory = keystonemiddleware.auth_token:filter_factory
|
||||
|
||||
# Prefix to prepend at the beginning of the path (string
|
||||
# value)
|
||||
@ -246,7 +246,7 @@ config file. For example in Nova, all middleware parameters can be removed
|
||||
from api-paste.ini::
|
||||
|
||||
[filter:authtoken]
|
||||
paste.filter_factory = keystoneclient.middleware.auth_token:filter_factory
|
||||
paste.filter_factory = keystonemiddleware.auth_token:filter_factory
|
||||
|
||||
and set in nova.conf::
|
||||
|
||||
@ -414,10 +414,10 @@ unsuccessful.
|
||||
Extended the request with additional User Information
|
||||
-----------------------------------------------------
|
||||
|
||||
:py:class:`keystoneclient.middleware.auth_token.AuthProtocol` extends the
|
||||
:py:class:`keystonemiddleware.auth_token.AuthProtocol` extends the
|
||||
request with additional information if the user has been authenticated. See the
|
||||
"What we add to the request for use by the OpenStack service" section in
|
||||
:py:mod:`keystoneclient.middleware.auth_token` for the list of fields set by
|
||||
:py:mod:`keystonemiddleware.auth_token` for the list of fields set by
|
||||
the auth_token middleware.
|
||||
|
||||
|
Before Width: | Height: | Size: 48 B After Width: | Height: | Size: 48 B |
Before Width: | Height: | Size: 3.7 KiB After Width: | Height: | Size: 3.7 KiB |
Before Width: | Height: | Size: 3.6 KiB After Width: | Height: | Size: 3.6 KiB |
@ -1,42 +0,0 @@
|
||||
Python bindings to the OpenStack Identity API (Keystone)
|
||||
========================================================
|
||||
|
||||
This is a client for OpenStack Identity API. There's a Python API for
|
||||
:doc:`Identity API v3 <using-api-v3>` and :doc:`v2 <using-api-v2>` (the
|
||||
:mod:`keystoneclient` modules), and a command-line script (installed as
|
||||
:doc:`keystone <man/keystone>`).
|
||||
|
||||
Contents:
|
||||
|
||||
.. toctree::
|
||||
:maxdepth: 1
|
||||
|
||||
man/keystone
|
||||
using-sessions
|
||||
using-api-v2
|
||||
using-api-v3
|
||||
|
||||
middlewarearchitecture
|
||||
|
||||
api/modules
|
||||
|
||||
Contributing
|
||||
============
|
||||
|
||||
Code is hosted `on GitHub`_. Submit bugs to the Keystone project on
|
||||
`Launchpad`_. Submit code to the ``openstack/python-keystoneclient`` project
|
||||
using `Gerrit`_.
|
||||
|
||||
.. _on GitHub: https://github.com/openstack/python-keystoneclient
|
||||
.. _Launchpad: https://launchpad.net/python-keystoneclient
|
||||
.. _Gerrit: http://wiki.openstack.org/GerritWorkflow
|
||||
|
||||
Run tests with ``python setup.py test``.
|
||||
|
||||
Indices and tables
|
||||
==================
|
||||
|
||||
* :ref:`genindex`
|
||||
* :ref:`modindex`
|
||||
* :ref:`search`
|
||||
|
@ -1,150 +0,0 @@
|
||||
========================================
|
||||
:program:`keystone` command line utility
|
||||
========================================
|
||||
|
||||
.. program:: keystone
|
||||
.. highlight:: bash
|
||||
|
||||
SYNOPSIS
|
||||
========
|
||||
|
||||
:program:`keystone` [options] <command> [command-options]
|
||||
|
||||
:program:`keystone help`
|
||||
|
||||
:program:`keystone help` <command>
|
||||
|
||||
|
||||
DESCRIPTION
|
||||
===========
|
||||
|
||||
The :program:`keystone` command line utility interacts with services providing
|
||||
OpenStack Identity API (e.g. Keystone).
|
||||
|
||||
To communicate with the API, you will need to be authenticated - and the
|
||||
:program:`keystone` provides multiple options for this.
|
||||
|
||||
While bootstrapping keystone the authentication is accomplished with a
|
||||
shared secret token and the location of the Identity API endpoint. The
|
||||
shared secret token is configured in keystone.conf as "admin_token".
|
||||
|
||||
You can specify those values on the command line with :option:`--os-token`
|
||||
and :option:`--os-endpoint`, or set them in environment variables:
|
||||
|
||||
.. envvar:: OS_SERVICE_TOKEN
|
||||
|
||||
Your keystone administrative token
|
||||
|
||||
.. envvar:: OS_SERVICE_ENDPOINT
|
||||
|
||||
Your Identity API endpoint
|
||||
|
||||
The command line options will override any environment variables set.
|
||||
|
||||
If you already have accounts, you can use your OpenStack username and
|
||||
password. You can do this with the :option:`--os-username`,
|
||||
:option:`--os-password`.
|
||||
|
||||
Keystone allows a user to be associated with one or more projects which are
|
||||
historically called tenants. To specify the project for which you want to
|
||||
authorize against, you may optionally specify a :option:`--os-tenant-id` or
|
||||
:option:`--os-tenant-name`.
|
||||
|
||||
Instead of using options, it is easier to just set them as environment
|
||||
variables:
|
||||
|
||||
.. envvar:: OS_USERNAME
|
||||
|
||||
Your Keystone username.
|
||||
|
||||
.. envvar:: OS_PASSWORD
|
||||
|
||||
Your Keystone password.
|
||||
|
||||
.. envvar:: OS_TENANT_NAME
|
||||
|
||||
Name of Keystone project.
|
||||
|
||||
.. envvar:: OS_TENANT_ID
|
||||
|
||||
ID of Keystone Tenant.
|
||||
|
||||
.. envvar:: OS_AUTH_URL
|
||||
|
||||
The OpenStack API server URL.
|
||||
|
||||
.. envvar:: OS_IDENTITY_API_VERSION
|
||||
|
||||
The OpenStack Identity API version.
|
||||
|
||||
.. envvar:: OS_CACERT
|
||||
|
||||
The location for the CA truststore (PEM formatted) for this client.
|
||||
|
||||
.. envvar:: OS_CERT
|
||||
|
||||
The location for the keystore (PEM formatted) containing the public
|
||||
key of this client. This keystore can also optionally contain the
|
||||
private key of this client.
|
||||
|
||||
.. envvar:: OS_KEY
|
||||
|
||||
The location for the keystore (PEM formatted) containing the private
|
||||
key of this client. This value can be empty if the private key is
|
||||
included in the OS_CERT file.
|
||||
|
||||
For example, in Bash you'd use::
|
||||
|
||||
export OS_USERNAME=yourname
|
||||
export OS_PASSWORD=yadayadayada
|
||||
export OS_TENANT_NAME=myproject
|
||||
export OS_AUTH_URL=http(s)://example.com:5000/v2.0/
|
||||
export OS_IDENTITY_API_VERSION=2.0
|
||||
export OS_CACERT=/etc/keystone/yourca.pem
|
||||
export OS_CERT=/etc/keystone/yourpublickey.pem
|
||||
export OS_KEY=/etc/keystone/yourprivatekey.pem
|
||||
|
||||
|
||||
OPTIONS
|
||||
=======
|
||||
|
||||
To get a list of available commands and options run::
|
||||
|
||||
keystone help
|
||||
|
||||
To get usage and options of a command::
|
||||
|
||||
keystone help <command>
|
||||
|
||||
|
||||
EXAMPLES
|
||||
========
|
||||
|
||||
Get information about endpoint-create command::
|
||||
|
||||
keystone help endpoint-create
|
||||
|
||||
View endpoints of OpenStack services::
|
||||
|
||||
keystone catalog
|
||||
|
||||
Create a 'service' project::
|
||||
|
||||
keystone tenant-create --name=service
|
||||
|
||||
Create service user for nova::
|
||||
|
||||
keystone user-create --name=nova \
|
||||
--tenant_id=<project ID> \
|
||||
--email=nova@nothing.com
|
||||
|
||||
View roles::
|
||||
|
||||
keystone role-list
|
||||
|
||||
|
||||
BUGS
|
||||
====
|
||||
|
||||
Keystone client is hosted in Launchpad so you can view current bugs at
|
||||
https://bugs.launchpad.net/python-keystoneclient/.
|
@ -1,117 +0,0 @@
|
||||
=================
|
||||
The Client v2 API
|
||||
=================
|
||||
|
||||
Introduction
|
||||
============
|
||||
|
||||
The main concepts in the Identity v2 API are:
|
||||
|
||||
* tenants
|
||||
* users
|
||||
* roles
|
||||
* services
|
||||
* endpoints
|
||||
|
||||
The client v2 API lets you query and make changes through
|
||||
managers. For example, to manipulate tenants, you interact with a
|
||||
``keystoneclient.v2_0.tenants.TenantManager`` object.
|
||||
|
||||
You obtain access to managers through via attributes of the
|
||||
``keystoneclient.v2_0.client.Client`` object. For example, the ``tenants``
|
||||
attribute of the ``Client`` class is a tenant manager::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> keystone.tenants.list() # List tenants
|
||||
|
||||
You create a valid ``keystoneclient.v2_0.client.Client`` object by passing
|
||||
authentication data to the constructor. Authentication and examples of common
|
||||
tasks are provided below.
|
||||
|
||||
You can generally expect that when the client needs to propagate an exception
|
||||
it will raise an instance of subclass of
|
||||
``keystoneclient.exceptions.ClientException``
|
||||
|
||||
Authenticating
|
||||
==============
|
||||
|
||||
There are two ways to authenticate against Keystone:
|
||||
* against the admin endpoint with the admin token
|
||||
* against the public endpoint with a username and password
|
||||
|
||||
If you are an administrator, you can authenticate by connecting to the admin
|
||||
endpoint and using the admin token (sometimes referred to as the service
|
||||
token). The token is specified as the ``admin_token`` configuration option in
|
||||
your keystone.conf config file, which is typically in /etc/keystone::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> token = '012345SECRET99TOKEN012345'
|
||||
>>> endpoint = 'http://192.168.206.130:35357/v2.0'
|
||||
>>> keystone = client.Client(token=token, endpoint=endpoint)
|
||||
|
||||
If you have a username and password, authentication is done against the
|
||||
public endpoint. You must also specify a tenant that is associated with the
|
||||
user::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> username='adminUser'
|
||||
>>> password='secreetword'
|
||||
>>> tenant_name='openstackDemo'
|
||||
>>> auth_url='http://192.168.206.130:5000/v2.0'
|
||||
>>> keystone = client.Client(username=username, password=password,
|
||||
... tenant_name=tenant_name, auth_url=auth_url)
|
||||
|
||||
Creating tenants
|
||||
================
|
||||
|
||||
This example will create a tenant named *openStackDemo*::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> keystone.tenants.create(tenant_name="openstackDemo",
|
||||
... description="Default Tenant", enabled=True)
|
||||
<Tenant {u'id': u'9b7962da6eb04745b477ae920ad55939', u'enabled': True, u'description': u'Default Tenant', u'name': u'openstackDemo'}>
|
||||
|
||||
Creating users
|
||||
==============
|
||||
|
||||
This example will create a user named *adminUser* with a password *secretword*
|
||||
in the opoenstackDemo tenant. We first need to retrieve the tenant::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> tenants = keystone.tenants.list()
|
||||
>>> my_tenant = [x for x in tenants if x.name=='openstackDemo'][0]
|
||||
>>> my_user = keystone.users.create(name="adminUser",
|
||||
... password="secretword",
|
||||
... tenant_id=my_tenant.id)
|
||||
|
||||
Creating roles and adding users
|
||||
===============================
|
||||
|
||||
This example will create an admin role and add the *my_user* user to that
|
||||
role, but only for the *my_tenant* tenant:
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> role = keystone.roles.create('admin')
|
||||
>>> my_tenant = ...
|
||||
>>> my_user = ...
|
||||
>>> keystone.roles.add_user_role(my_user, role, my_tenant)
|
||||
|
||||
Creating services and endpoints
|
||||
===============================
|
||||
|
||||
This example will create the service and corresponding endpoint for the
|
||||
Compute service::
|
||||
|
||||
>>> from keystoneclient.v2_0 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> service = keystone.services.create(name="nova", service_type="compute",
|
||||
... description="Nova Compute Service")
|
||||
>>> keystone.endpoints.create(
|
||||
... region="RegionOne", service_id=service.id,
|
||||
... publicurl="http://192.168.206.130:8774/v2/%(tenant_id)s",
|
||||
... adminurl="http://192.168.206.130:8774/v2/%(tenant_id)s",
|
||||
... internalurl="http://192.168.206.130:8774/v2/%(tenant_id)s")
|
@ -1,113 +0,0 @@
|
||||
=================
|
||||
The Client v3 API
|
||||
=================
|
||||
|
||||
Introduction
|
||||
============
|
||||
|
||||
The main concepts in the Identity v3 API are:
|
||||
|
||||
* credentials
|
||||
* domains
|
||||
* endpoints
|
||||
* groups
|
||||
* policies
|
||||
* projects
|
||||
* role assignments
|
||||
* roles
|
||||
* services
|
||||
* trusts
|
||||
* users
|
||||
|
||||
The :py:mod:`keystoneclient.v3.client` API lets you query and make changes
|
||||
through ``managers``. For example, to manipulate a project (formerly
|
||||
called tenant), you interact with a
|
||||
:py:class:`keystoneclient.v3.projects.ProjectManager` object.
|
||||
|
||||
You obtain access to managers through attributes of a
|
||||
:py:class:`keystoneclient.v3.client.Client` object. For example, the
|
||||
``projects`` attribute of a ``Client`` object is a projects manager::
|
||||
|
||||
>>> from keystoneclient.v3 import client
|
||||
>>> keystone = client.Client(...)
|
||||
>>> keystone.projects.list() # List projects
|
||||
|
||||
While it is possible to instantiate a
|
||||
:py:class:`keystoneclient.v3.client.Client` object (as done above for
|
||||
clarity), the recommended approach is to use the discovery mechanism
|
||||
provided by the :py:class:`keystoneclient.client.Client` class. The
|
||||
appropriate class will be instantiated depending on the API versions
|
||||
available::
|
||||
|
||||
>>> from keystoneclient import client
|
||||
>>> keystone =
|
||||
... client.Client(auth_url='http://localhost:5000', ...)
|
||||
>>> type(keystone)
|
||||
<class 'keystoneclient.v3.client.Client'>
|
||||
|
||||
One can force the use of a specific version of the API, either by
|
||||
using the ``version`` keyword argument::
|
||||
|
||||
>>> from keystoneclient import client
|
||||
>>> keystone = client.Client(auth_url='http://localhost:5000',
|
||||
version=(2,), ...)
|
||||
>>> type(keystone)
|
||||
<class 'keystoneclient.v2_0.client.Client'>
|
||||
>>> keystone = client.Client(auth_url='http://localhost:5000',
|
||||
version=(3,), ...)
|
||||
>>> type(keystone)
|
||||
<class 'keystoneclient.v3.client.Client'>
|
||||
|
||||
Or by specifying directly the specific API version authentication URL
|
||||
as the auth_url keyword argument::
|
||||
|
||||
>>> from keystoneclient import client
|
||||
>>> keystone =
|
||||
... client.Client(auth_url='http://localhost:5000/v2.0', ...)
|
||||
>>> type(keystone)
|
||||
<class 'keystoneclient.v2_0.client.Client'>
|
||||
>>> keystone =
|
||||
... client.Client(auth_url='http://localhost:5000/v3', ...)
|
||||
>>> type(keystone)
|
||||
<class 'keystoneclient.v3.client.Client'>
|
||||
|
||||
Upon successful authentication, a :py:class:`keystoneclient.v3.client.Client`
|
||||
object is returned (when using the Identity v3 API). Authentication and
|
||||
examples of common tasks are provided below.
|
||||
|
||||
You can generally expect that when the client needs to propagate an
|
||||
exception it will raise an instance of subclass of
|
||||
``keystoneclient.exceptions.ClientException`` (see
|
||||
:py:class:`keystoneclient.openstack.common.apiclient.exceptions.ClientException`)
|
||||
|
||||
Authenticating
|
||||
==============
|
||||
|
||||
You can authenticate against Keystone using a username, a user domain
|
||||
name (which will default to 'Default' if it is not specified) and a
|
||||
password::
|
||||
|
||||
>>> from keystoneclient import client
|
||||
>>> auth_url = 'http://localhost:5000'
|
||||
>>> username = 'adminUser'
|
||||
>>> user_domain_name = 'Default'
|
||||
>>> password = 'secreetword'
|
||||
>>> keystone = client.Client(auth_url=auth_url, version=(3,),
|
||||
... username=username, password=password,
|
||||
... user_domain_name=user_domain_name)
|
||||
|
||||
You may optionally specify a domain or project (along with its project
|
||||
domain name), to obtain a scoped token::
|
||||
|
||||
>>> from keystoneclient import client
|
||||
>>> auth_url = 'http://localhost:5000'
|
||||
>>> username = 'adminUser'
|
||||
>>> user_domain_name = 'Default'
|
||||
>>> project_name = 'demo'
|
||||
>>> project_domain_name = 'Default'
|
||||
>>> password = 'secreetword'
|
||||
>>> keystone = client.Client(auth_url=auth_url, version=(3,),
|
||||
... username=username, password=password,
|
||||
... user_domain_name=user_domain_name,
|
||||
... project_name=project_name,
|
||||
... project_domain_name=project_domain_name)
|
@ -1,198 +0,0 @@
|
||||
==============
|
||||
Using Sessions
|
||||
==============
|
||||
|
||||
Introduction
|
||||
============
|
||||
|
||||
The :py:class:`keystoneclient.session.Session` class was introduced into
|
||||
keystoneclient as an attempt to bring a unified interface to the various
|
||||
OpenStack clients that share common authentication and request parameters
|
||||
between a variety of services.
|
||||
|
||||
The model for using a Session and auth plugin as well as the general terms used
|
||||
have been heavily inspired by the `requests <http://docs.python-requests.org>`_
|
||||
library. However neither the Session class nor any of the authentication
|
||||
plugins rely directly on those concepts from the requests library so you should
|
||||
not expect a direct translation.
|
||||
|
||||
Features
|
||||
--------
|
||||
|
||||
- Common client authentication
|
||||
|
||||
Authentication is handled by one of a variety of authentication plugins and
|
||||
then this authentication information is shared between all the services that
|
||||
use the same Session object.
|
||||
|
||||
- Security maintenance
|
||||
|
||||
Security code is maintained in a single place and reused between all
|
||||
clients such that in the event of problems it can be fixed in a single
|
||||
location.
|
||||
|
||||
- Standard discovery mechanisms
|
||||
|
||||
Clients are not expected to have any knowledge of an identity token or any
|
||||
other form of identification credential. Service and endpoint discovery are
|
||||
handled by the Session and plugins.
|
||||
|
||||
|
||||
Sessions for Users
|
||||
==================
|
||||
|
||||
The Session object is the contact point to your OpenStack cloud services. It
|
||||
stores the authentication credentials and connection information required to
|
||||
communicate with OpenStack such that it can be reused to communicate with many
|
||||
services. When creating services this Session object is passed to the client
|
||||
so that it may use this information.
|
||||
|
||||
A Session will authenticate on demand. When a request that requires
|
||||
authentication passes through the Session the authentication plugin will be
|
||||
asked for a valid token. If a valid token is available it will be used
|
||||
otherwise the authentication plugin may attempt to contact the authentication
|
||||
service and fetch a new one.
|
||||
|
||||
An example from keystoneclient::
|
||||
|
||||
>>> from keystoneclient.auth.identity import v3
|
||||
>>> from keystoneclient import session
|
||||
>>> from keystoneclient.v3 import client
|
||||
|
||||
>>> auth = v3.Password(auth_url='https://my.keystone.com:5000/v2.0',
|
||||
... username='myuser',
|
||||
... password='mypassword',
|
||||
... project_id='proj')
|
||||
>>> sess = session.Session(auth=auth,
|
||||
... verify='/path/to/ca.cert')
|
||||
>>> ks = client.Client(session=sess)
|
||||
>>> users = ks.users.list()
|
||||
|
||||
As clients adopt this means of operating they will be created in a similar
|
||||
fashion by passing the Session object to the client's constructor.
|
||||
|
||||
|
||||
Migrating keystoneclient to use a Session
|
||||
-----------------------------------------
|
||||
|
||||
By using a session with a keystonclient Client we define that you have opted in
|
||||
to new behaviour defined by the session. For example authentication is now
|
||||
on-demand rather than on creation. To allow this change in behaviour there are
|
||||
a number of functions that have changed behaviour or are no longer available.
|
||||
|
||||
For example the
|
||||
:py:meth:`keystoneclient.httpclient.HTTPClient.authenticate` command used
|
||||
to be able to always re-authenticate the current client and fetch a new token.
|
||||
As this is now controlled by the Session and not the client this has changed,
|
||||
however the function will still exist to provide compatibility with older
|
||||
clients.
|
||||
|
||||
Likewise certain parameters such as ``user_id`` and ``auth_token`` that used to
|
||||
be available on the client object post authentication will remain
|
||||
uninitialized.
|
||||
|
||||
When converting an application to use a session object with keystoneclient you
|
||||
should be aware of the possibility of changes to authentication and
|
||||
authentication parameters and make sure to test your code thoroughly. It should
|
||||
have no impact on the typical CRUD interaction with the client.
|
||||
|
||||
|
||||
Sharing Authentication Plugins
|
||||
------------------------------
|
||||
|
||||
A session can only contain one authentication plugin however there is nothing
|
||||
that specifically binds the authentication plugin to that session, a new
|
||||
Session can be created that reuses the existing authentication plugin::
|
||||
|
||||
>>> new_sess = session.Session(auth=sess.auth,
|
||||
verify='/path/to/different-cas.cert')
|
||||
|
||||
In this case we cannot know which session object will be used when the plugin
|
||||
performs the authentication call so the command must be able to succeed with
|
||||
either.
|
||||
|
||||
Authentication plugins can also be provided on a per-request basis. This will
|
||||
be beneficial in a situation where a single session is juggling multiple
|
||||
authentication credentials::
|
||||
|
||||
>>> sess.get('https://my.keystone.com:5000/v3',
|
||||
auth=my_auth_plugin)
|
||||
|
||||
If an auth plugin is provided via parameter then it will override any auth
|
||||
plugin on the session.
|
||||
|
||||
Sessions for Client Developers
|
||||
==============================
|
||||
|
||||
Sessions are intended to take away much of the hassle of dealing with
|
||||
authentication data and token formats. Clients should be able to specify filter
|
||||
parameters for selecting the endpoint and have the parsing of the catalog
|
||||
managed for them.
|
||||
|
||||
Authentication
|
||||
--------------
|
||||
|
||||
When making a request with a session object you can simply pass the keyword
|
||||
parameter ``authenticated`` to indicate whether the argument should contain a
|
||||
token, by default a token is included if an authentication plugin is available::
|
||||
|
||||
>>> # In keystone this route is unprotected by default
|
||||
>>> resp = sess.get('https://my.keystone.com:5000/v3',
|
||||
authenticated=False)
|
||||
|
||||
|
||||
Service Discovery
|
||||
-----------------
|
||||
|
||||
In OpenStack the URLs of available services are distributed to the user as a
|
||||
part of the token they receive called the Service Catalog. Clients are expected
|
||||
to use the URLs from the Service Catalog rather than have them provided.
|
||||
|
||||
In general a client does not need to know the full URL for the server that they
|
||||
are communicating with, simply that it should send a request to a path
|
||||
belonging to the correct service.
|
||||
|
||||
This is controlled by the ``endpoint_filter`` parameter to a request which
|
||||
contains all the information an authentication plugin requires to determine the
|
||||
correct URL to which to send a request. When using this mode only the path for
|
||||
the request needs to be specified::
|
||||
|
||||
>>> resp = session.get('/v3/users',
|
||||
endpoint_filter={'service_type': 'identity',
|
||||
'interface': 'public',
|
||||
'region_name': 'myregion'})
|
||||
|
||||
``endpoint_filter`` accepts a number of arguments with which it can determine
|
||||
an endpoint url:
|
||||
|
||||
- ``service_type``: the type of service. For example ``identity``, ``compute``,
|
||||
``volume`` or many other predefined identifiers.
|
||||
|
||||
- ``interface``: the network exposure the interface has. This will be one of:
|
||||
|
||||
- ``public``: An endpoint that is available to the wider internet or network.
|
||||
- ``internal``: An endpoint that is only accessible within the private network.
|
||||
- ``admin``: An endpoint to be used for administrative tasks.
|
||||
|
||||
- ``region_name``: the name of the region where the endpoint resides.
|
||||
|
||||
The endpoint filter is a simple key-value filter and can be provided with any
|
||||
number of arguments. It is then up to the auth plugin to correctly use the
|
||||
parameters it understands.
|
||||
|
||||
The session object determines the URL matching the filter and append to it the
|
||||
provided path and so create a valid request. If multiple URL matches are found
|
||||
then any one may be chosen.
|
||||
|
||||
While authentication plugins will endeavour to maintain a consistent set of
|
||||
arguments for an ``endpoint_filter`` the concept of an authentication plugin is
|
||||
purposefully generic and a specific mechanism may not know how to interpret
|
||||
certain arguments and ignore them. For example the
|
||||
:py:class:`keystoneclient.auth.token_endpoint.Token` plugin (which is used when
|
||||
you want to always use a specific endpoint and token combination) will always
|
||||
return the same endpoint regardless of the parameters to ``endpoint_filter`` or
|
||||
a custom OpenStack authentication mechanism may not have the concept of
|
||||
multiple ``interface`` options and choose to ignore that parameter.
|
||||
|
||||
There is some expectation on the user that they understand the limitations of
|
||||
the authentication system they are using.
|