Merge "Minor updates to rbac doc"

This commit is contained in:
Zuul 2022-03-03 00:23:17 +00:00 committed by Gerrit Code Review
commit 8eec87bbe4
1 changed files with 9 additions and 9 deletions

View File

@ -16,7 +16,7 @@ customization of these policies to consult some reference material
in hopes of understanding the context.
* `Keystone Adminstrator Guide - Service API Protection <https://docs.openstack.org/keystone/latest/admin/service-api-protection.html>`_
* `Ironic Scoped Role Based Access Control Specification <https://specs.openstack.org/openstack/ironic-specs/specs/not-implemented/secure-rbac.html>`_
* `Ironic Scoped Role Based Access Control Specification <https://specs.openstack.org/openstack/ironic-specs/specs/17.0/secure-rbac.html>`_
Historical Context - How we reached our access model
----------------------------------------------------
@ -45,7 +45,7 @@ policy enforcement framework the information necessary to make decisions.
System scoped requests very much align with the access controls of Ironic
before the Secure RBAC effort. The original custom role ``baremetal_admin``
privileges are identical to a system scoped ``admin``'s privileges.
Similarly ``baremetal_reader`` is identical to a system scoped ``reader``.
Similarly ``baremetal_observer`` is identical to a system scoped ``reader``.
In these concepts, the ``admin`` is allowed to create/delete objects/items.
The ``reader`` is allowed to read details about items and is intended for
users who may need an account with read-only access for or front-line support
@ -101,7 +101,7 @@ How Project Scoped Works
------------------------
Ironic has two project use models where access is generally more delegative
to an ``owner`` where access to a ``lessee`` is generally more utilitarian.
to an ``owner`` and access to a ``lessee`` is generally more utilitarian.
The purpose of an owner, is more to enable the System Operator to delegate
much of the administrative activity of a Node to the owner.
@ -131,13 +131,13 @@ Field value visibility restrictions
Ironic's API, by default has a concept of filtering node values to prevent
sensitive data from being leaked. System scoped users are subjected to basic
restrictions, where as project scoped users are, by default, examined further
restrictions, whereas project scoped users are, by default, examined further
and against additional policies. This threshold is controlled with the
``baremetal:node:get:filter_threshold``.
By default, the following fields are masked on Nodes and are controlled by the
associated policies. By default, owner's are able to see insight into the
infrastructure, where as lessee users *CANNOT* view these fields by default.
associated policies. By default, owners are able to see insight into the
infrastructure, whereas lessee users *CANNOT* view these fields by default.
* ``last_error`` - ``baremetal:node:get:last_error``
* ``reservation`` - ``baremetal:node:get:reservation``
@ -175,7 +175,7 @@ Allocations
~~~~~~~~~~~
The ``allocations`` endpoint of the API is somewhat different than other
other endpoints as it allows for the allocation of physical machines to
endpoints as it allows for the allocation of physical machines to
an admin. In this context, there is not already an ``owner`` or ``project_id``
to leverage to control access for the creation process, any project member
does have the inherent privilege of requesting an allocation. That being said,
@ -188,7 +188,7 @@ and any new allocation being requested with a specific owner, if made in
the allocation.
Ultimately, an operational behavior difference exists between the ``owner``
and ``lessee`` rights in terms of allocations exists. With the standard
and ``lessee`` rights in terms of allocations. With the standard
access rights, ``lessee`` users are able to create allocations if they
own nodes which are not allocated or deployed, but they cannot reprovision
nodes when using only a ``member`` role. This limitation is not the case
@ -216,7 +216,7 @@ the user to be a ``system`` scoped user with ``admin`` privileges.
The most noticeable difference for API consumers is the HTTP 403 access
code is now mainly a HTTP 404 access code. The access concept has changed
from "Does the user user broadly has access to the API?" to
from "Does the user broadly have access to the API?" to
"Does user have access to the node, and then do they have access
to the specific resource?".