docs: Add Patrole RBAC field guide
This documentation update adds a README.rst to
patrole_tempest_plugin/tests/api which explains the RBAC field
guide. It is modeled after Tempest's API field guide [0][1]. The
README.rst is then referenced in a new field_guide section
under doc/source/field_guide.
[0] https://github.com/openstack/tempest/tree/28b252f7f61a2dfc585089265cdb3118defbcfff/tempest/api
[1] https://docs.openstack.org/tempest/latest/field_guide/api.html#api-field-guide
Change-Id: I877ce4a1f681bd483c7f71b02fd7bb2b4d3b3e2a
diff --git a/doc/source/field_guide/index.rst b/doc/source/field_guide/index.rst
new file mode 100644
index 0000000..ba06c42
--- /dev/null
+++ b/doc/source/field_guide/index.rst
@@ -0,0 +1,60 @@
+.. _patrole-field-guide:
+
+============================
+Patrole Field Guide Overview
+============================
+
+Testing Scope
+=============
+
+Patrole testing scope is strictly confined to Role-Based Access Control
+(RBAC). In OpenStack, ``oslo.policy`` is the RBAC library used by all
+major services. Thus, Patrole is concerned with validating that public API
+endpoints are correctly using ``oslo.policy`` for authorization.
+
+In other words, all tests in Patrole are RBAC tests.
+
+:ref:`rbac_field_guide`
+=======================
+
+RBAC tests are `Tempest`_-like API tests plus Patrole's
+:ref:`rbac-validation`. All Patrole tests are RBAC validation tests for the
+OpenStack API.
+
+.. _Tempest: https://docs.openstack.org/tempest/latest/
+
+Stable Tests
+============
+
+In the discussion below, "correct" means that a test is consistent with
+a service's API-to-policy mapping and "stable" means that a test should
+require minimal maintenance for the supported releases.
+
+Present
+-------
+
+During the Queens release, a `governance spec`_ was pushed to support policy
+in code, which documents the mapping between APIs and each of their policies.
+
+This documentation is an important prerequisite for ensuring that Patrole
+tests for a given service are correct. This mapping can be referenced to
+confirm that Patrole's assumed mapping for a test is correct. For
+example, Nova has implemented policy in code which can be used to verify
+that Patrole's Nova RBAC tests use the same mapping.
+
+If a given service does not have policy in code, this implies that it is
+*more likely* that the RBAC tests for that service are inconsistent with the
+*intended* policy mapping. Until that service implements policy in code, it
+is difficult for Patrole maintainers to verify that tests for that service
+are correct.
+
+Future
+------
+
+Once all services that Patrole tests have implemented policy in code --
+and once Patrole has updated all its tests in accordance with the policy in
+code documentation -- then Patrole tests can guaranteed to be stable.
+
+This stability will be denoted with a 1.0 version release.
+
+.. _governance spec: https://governance.openstack.org/tc/goals/queens/policy-in-code.html
diff --git a/doc/source/field_guide/rbac.rst b/doc/source/field_guide/rbac.rst
new file mode 100644
index 0000000..2654d31
--- /dev/null
+++ b/doc/source/field_guide/rbac.rst
@@ -0,0 +1,71 @@
+.. _rbac_field_guide:
+
+Patrole Field Guide to RBAC Tests
+=================================
+
+
+What are these tests?
+---------------------
+
+Patrole's primary responsibility is to ensure that your OpenStack cloud
+has properly configured Role-Based Access Control (RBAC). All Patrole
+tests cases are devoted to this responsibility. Tempest API clients
+and utility functions are leveraged to accomplish this goal, but such
+functionality is secondary to RBAC validation.
+
+Like Tempest, Patrole not only tests expected positive paths for RBAC
+validation, but also -- and more importantly -- negative paths. While
+Patrole could be thought of as validating RBAC, it more importantly
+verifies that your OpenStack cloud is secure from the perspective of
+RBAC (there are many gotchas when it comes to security, not just RBAC).
+
+Negative paths are arguably more important than positive paths when it
+comes to RBAC and by extension security, because it is essential that
+your cloud be secure from unauthorized access. For example, while it is
+important to verify that the admin role has access to admin-level
+functionality, it is of critical importance to verify that non-admin roles
+*do not* have access to such functionality.
+
+Unlike Tempest, Patrole accomplishes negative testing implicitly -- by
+abstracting it away in the background. Patrole dynamically determines
+whether a role should have access to an API depending on your cloud's
+policy configuration and then confirms whether that is true or false.
+
+
+Why are these tests in Patrole?
+-------------------------------
+
+These tests constitute the core mission in Patrole: to verify RBAC. These
+tests are mainly intended to validate RBAC, but can also *unofficially*
+be used to discover the policy-to-API mapping for an OpenStack component.
+
+It could be argued that some of these tests could be implemented in
+the projects themselves, but that approach has the following shortcomings:
+
+* The projects do not validate RBAC from an integration testing perspective.
+* By extension, RBAC across cross-service communication is not usually
+ validated.
+* The projects' tests do not pass all the metadata to ``oslo.policy`` that is
+ in reality passed by the deployed server to that library to determine
+ whether a given user is authorized to perform an API action.
+* The projects do not exhaustively do RBAC testing for all positive and
+ negative paths.
+* Patrole is designed to work with any role via configuration settings, but
+ on the other hand the projects handpick which roles to test.
+
+
+Scope of these tests
+--------------------
+
+RBAC tests should always use the Tempest implementation of the
+OpenStack API, to take advantage of Tempest's stable library.
+
+Each test should test a specific API endpoint and the related policy.
+
+Each policy should be tested in isolation of one another -- or at least
+as close to this rule as possible -- to ensure proper validation of RBAC.
+
+Each test should be able to work for positive and negative paths.
+
+All tests should be able to be run on their own, not depending on the
+state created by a previous test.