Adding ansible_base.api_documentation
to the INSTALL_APPS which extends the schema
to include an LLM-friendly description
to each endpoint
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Peter Braun <pbraun@redhat.com>
Modify the invocation of @task_awx to accept timeout and
on_duplicate keyword arguments. These arguments are
only used in the new dispatcher implementation.
Add decorator params:
- timeout
- on_duplicate
to tasks to ensure better recovery for
stuck or long-running processes.
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
settings.SUBSCRIPTIONS_USERNAME and
settings.SUBSCRIPTIONS_CLIENT_ID
should be mutually exclusive. This is because
the POST to api/v2/config/attach/ accepts only
a subscription_id, and infers which credentials to
use based on settings. If both are set, it is ambiguous
and can lead to unexpected 400s when attempting
to attach a license.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Allow users to do subscription management using
Red Hat username and password.
In basic auth case, the candlepin API
at subscriptions.rhsm.redhat.com will be used instead
of console.redhat.com.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Syncing from new rbac to old rbac locally calls the
disable_rbac_sync() context manager.
If rbac sync is disabled, we do not need to remote
sync, as we can assume the remote syncing already
occurred in the viewset.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Remove Controller specific roles
Removes
- Controller Organization Admin
- Controller Organization Member
- Controller Team Admin
- Controller Team Member
- Controller System Auditor
Going forward the platform role definitions
will be used, e.g. Organization Member
The migration will take care of any assignments
with those controller specific roles and use
the platform roles instead.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
When POSTing to console.redhat.com, fallback
to using basic auth method if OAUTH via
service accounts fails
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fallback to basic auth if OAUTH to console.redhat.com fails
Notes:
Envoy has a timeout of 30 seconds, so
the total timeout should be less than that.
(5, 20) means 5 seconds to connect to server,
20 seconds to start reading data.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Ensure service account authentication is being used
when falling back to using SUBSCRIPTIONS_CLIENT_ID.
Additional change:
Subscription data can return two types of capacities:
Sockets and Nodes
For determining overall capacity
if capacity name is Nodes:
capacity quantity x subscription quantity
if capacity name is Sockets:
capacity quantity / 2 (minimum of 1) x subscription quantity
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
* Update subscription API to use service accounts
Update code to pull subscriptions from
console.redhat.com instead of
subscription.rhsm.redhat.com
Uses service account client ID and client secret
instead of username/password, which is being
deprecated in July 2025.
Additional changes:
- In awx.awx.subscriptions module, use new service
account params rather than old basic auth params
- Update awx.awx.license module to use subscription_id
instead of pool_id. This is due to using a different API,
which identifies unique subscriptions by subscriptionID
instead of pool ID.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Chris Meyers <chris.meyers.fsu@gmail.com>
Co-authored-by: Peter Braun <pbraun@redhat.com>
* fix token name
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
* Fix Subscriptions credentials fallback
Ensure service account authentication is being used
when falling back to using SUBSCRIPTIONS_CLIENT_ID.
Additional change:
Subscription data can return two types of capacities:
Sockets and Nodes
For determining overall capacity
if capacity name is Nodes:
capacity quantity x subscription quantity
if capacity name is Sockets:
capacity quantity / 2 (minimum of 1) x subscription quantity
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Chris Meyers <chris.meyers.fsu@gmail.com>
Co-authored-by: Peter Braun <pbraun@redhat.com>
Update code to pull subscriptions from
console.redhat.com instead of
subscription.rhsm.redhat.com
Uses service account client ID and client secret
instead of username/password, which is being
deprecated in July 2025.
Additional changes:
- In awx.awx.subscriptions module, use new service
account params rather than old basic auth params
- Update awx.awx.license module to use subscription_id
instead of pool_id. This is due to using a different API,
which identifies unique subscriptions by subscriptionID
instead of pool ID.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Chris Meyers <chris.meyers.fsu@gmail.com>
Co-authored-by: Peter Braun <pbraun@redhat.com>
Modifies to_container_path to accept an optional
container_root parameter.
Normally this defaults to /runner, but in K8S
environments, project updates run from the
private_data_dir, e.g. /tmp/awx_1_123abc, not
/runner.
In that situation, we just pass in private_data_dir
as the container_root.
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: TVo <thavo@redhat.com>
In OCP/K8S, projects run in the task pod's ee container. The private_data_dir is not extracted to /runner. Instead, the project update runs directly from the mounted in private_data_dir, e.g. /tmp/awx_1_abcd.
When injecting a credential that uses extra vars, we pass the private_data_dir as as the container_root, so that the correct command line argument is generated, e.g. "-e /tmp/awx_1_abcd/env/extra_var_file".
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fixes an issue where schedules were not running at
the correct time.
Details:
DST is Daylights Saving Time
If the rrule dtstart is "in" a DST period (i.e., March to November)
and the current date is outside of the DST, then the fast forwarding
is not correct.
This is because datetime timedeltas do not honor DST boundaries
The Fix:
Convert the rrule dtstart to UTC before doing operations. Then,
convert back to the original timezone at the end.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fixes an issue where schedules were not running at
the correct time.
Details:
DST is Daylights Saving Time
If the rrule dtstart is "in" a DST period (i.e., March to November)
and the current date is outside of the DST, then the fast forwarding
is not correct.
This is because datetime timedeltas do not honor DST boundaries
The Fix:
Convert the rrule dtstart to UTC before doing operations. Then,
convert back to the original timezone at the end.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fixes a bug where a schedule that was created
to run only once will continue to run repeatedly.
e.g. an rrule with
dtstart 20240730; count 1; freq MINUTELY
This job will run on 20240730, and should never
run again.
However, the next time the schedule
update_computed_fields runs, the dtstart
will fast forward to today's date, and
next_run will be computed from that. This will trigger
the job to run again, which is not intended.
If count is set, we just should not fast forward the
rrule and always calculate next_run based on original
dtstart.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fixes a bug where a schedule that was created
to run only once will continue to run repeatedly.
e.g. an rrule with
dtstart 20240730; count 1; freq MINUTELY
This job will run on 20240730, and should never
run again.
However, the next time the schedule
update_computed_fields runs, the dtstart
will fast forward to today's date, and
next_run will be computed from that. This will trigger
the job to run again, which is not intended.
If count is set, we just should not fast forward the
rrule and always calculate next_run based on original
dtstart.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
By stable, we mean future occurrences of the rrule
should be the same before and after the fast forward
operation.
The problem before was that we were fast forwarding to
7 days ago. For some rrules, this does not retain the old
occurrences. Thus, jobs would launch at unexpected times.
This change makes sure we fast forward in increments of
the rrule INTERVAL, thus the new dtstart should be in the
occurrence list of the old rrule.
Additionally, code is updated to fast forward
EXRULE (exclusion rules) in addition to RRULE
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: 🇺🇦 Sviatoslav Sydorenko (Святослав Сидоренко) <wk.cvs.github@sydorenko.org.ua>
By stable, we mean future occurrences of the rrule
should be the same before and after the fast forward
operation.
The problem before was that we were fast forwarding to
7 days ago. For some rrules, this does not retain the old
occurrences. Thus, jobs would launch at unexpected times.
This change makes sure we fast forward in increments of
the rrule INTERVAL, thus the new dtstart should be in the
occurrence list of the old rrule.
Additionally, code is updated to fast forward
EXRULE (exclusion rules) in addition to RRULE
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: 🇺🇦 Sviatoslav Sydorenko (Святослав Сидоренко) <wk.cvs.github@sydorenko.org.ua>
Adding credential and execution environment roles
validates that the user belongs to the same org
as the credential or EE.
In some situations, the user-org membership has not
yet been synced from gateway to controller.
In this case, controller will make a request to
gateway to check if the user is part of the org.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Adding credential and execution environment roles
validates that the user belongs to the same org
as the credential or EE.
In some situations, the user-org membership has not
yet been synced from gateway to controller.
In this case, controller will make a request to
gateway to check if the user is part of the org.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
User and Team assignments using the DAB
RBAC system will be translated back to the old
Role system.
This ensures better backward compatibility and
addresses some inconsistences in the UI that were
relying on older RBAC endpoints.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Alan Rominger <arominge@redhat.com>
This is to emphasize that this role is specific
to controller component. That is, not an auditor
for the entire AAP platform.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
User and Team assignments using the DAB
RBAC system will be translated back to the old
Role system.
This ensures better backward compatibility and
addresses some inconsistences in the UI that were
relying on older RBAC endpoints.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Alan Rominger <arominge@redhat.com>
Fixes bug where creating a new user will
request a new awx_sessionid cookie, invalidating
the previous session.
Do not refresh session if updating or
creating a password for a different user.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
This is to emphasize that this role is specific
to controller component. That is, not an auditor
for the entire AAP platform.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Adds the following managed Role Definitions
Controller Team Admin
Controller Team Member
Controller Organization Admin
Controller Organization Member
These have the same permission set as the
platform roles (without the Controller prefix)
Adding members to teams and orgs via the legacy RBAC system
will use these role definitions.
Other changes:
- Bump DAB to 2024.08.22
- Set ALLOW_LOCAL_ASSIGNING_JWT_ROLES to False in defaults.py.
This setting prevents assignments to the platform roles (e.g. Team Member).
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Adds the following managed Role Definitions
Controller Team Admin
Controller Team Member
Controller Organization Admin
Controller Organization Member
These have the same permission set as the
platform roles (without the Controller prefix)
Adding members to teams and orgs via the legacy RBAC system
will use these role definitions.
Other changes:
- Bump DAB to 2024.08.22
- Set ALLOW_LOCAL_ASSIGNING_JWT_ROLES to False in defaults.py.
This setting prevents assignments to the platform roles (e.g. Team Member).
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Fixes bug where creating a new user will
request a new awx_sessionid cookie, invalidating
the previous session.
Do not refresh session if updating or
creating a password for a different user.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Change django url dispatcher to serve up ui_next files instead of old ui files
Old UI will not be served with this change
Github CI still runs old ui tests (to be removed in another PR)
Remove the Github workflows that build old UI
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Old RBAC system hits DOESNOTEXIST query errors
if a user deletes an org while a workflow job is active.
The error is triggered by
1. starting workflow job
2. delete the org that the workflow job is a part of
3. The workflow changes status (e.g. pending to waiting)
This error message would surface
awx.main.models.rbac.Role.DoesNotExist: Role matching
query does not exist.
The fix is wrap the query in a try catch, and skip
over some logic if the roles don't exist.
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
A user needs to be a member of the org
in order to use a credential in that org.
We were incorrectly checking for "change"
permission of the org, instead of "member".
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Validate role assignment if org defined
Check that organization is defined on credential
before running queries.
Fixes a "None type does not have attribute id" error.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Utilizes the `validate_role_assignment` callback
from dab (see dab PR #490) to prevent granting credential
access to a user of another organization.
This logic will work for role_user_assignments
and role_team_assignments endpoints.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Script was falsely identifying cross-linked
parents. It needs to check if parent roles if
content type is Team and role_field is
member_role OR admin_role.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
rename AWX_DIRECT_SHARED_RESOURCE_MANAGEMENT_ENABLED
to
ALLOW_LOCAL_RESOURCE_MANAGEMENT
- clearer meaning
- drop prefix so the same setting is used across the platform
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
If more than one schedule for a unified job template
is removed at once, a race condition can arise.
example scenario: delete schedules with ids 7 and 8
- unified job template next_schedule is currently 7
- on delete of schedule 7, update_computed_fields will try to set
next_schedule to 8
- but while this logic is occurring, another transaction
is deleting 8
This leads to a db IntegrityError
The solution here is to call select_for_update() on the
next schedule, so that 8 cannot be deleted until
the transaction for deleting 7 is completed.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
PG_TLS=true make docker-compose
This will add some extra startup commands
for the postgres container to generate a key and
cert to use for postgres connections.
It will also mount in pgssl.conf which has ssl configuration.
This can be useful for debugging issues that only surface
when using ssl postgres connections.
* Prevent modifying shared resources
Adds a class decorator to prevent modifying shared resources
when gateway is being used.
AWX_DIRECT_SHARED_RESOURCE_MANAGEMENT_ENABLED is the setting
to enable/disable this feature.
Works by overriding these view methods:
- create
- delete
- perform_update
create and delete are overridden to raise a
PermissionDenied exception.
perform_update is overridden to check if any shared
fields are being modified, and raise a PermissionDenied
exception if so.
Additional changes:
Prevent sso conf from registering external authentication related settings if
AWX_DIRECT_SHARED_RESOURCE_MANAGEMENT_ENABLED is False
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Co-authored-by: Hao Liu <44379968+TheRealHaoLiu@users.noreply.github.com>
This change makes "wait: true" for jobs and syncs
look at the event_processing_finished instead of
finished field.
Right now there is a race condition where
a module might try to delete an inventory, but the events
for an inventory sync have not yet finished. We have a
RelatedJobsPreventDeleteMixin that checks for this condition.
bulk jobs don't have event_processing_finished so we just
use finished field in that case.
openssl 3.2.0 has incompatiblity issues with
the libpq version we are using, and causes
some C runtime errors:
"double free or corruption (out)"
see awx issue #15136
also this issue
github.com/conan-io/conan-center-index/pull/22615
once the libpq libraries on centos stream9 are
updated with the patch, we can unpin openssl
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Currently the association box displays a
list of available instances/addresses that can
be peered to.
The pagination issue arises as follows:
- fetch 5 instances (based on page_size)
- filter these instances down based on some
criteria (like is_internal: false)
- show results
Filtering down the results inside of the fetch
method results in pagnation errors (may return fewer than
5, for example)
instead, do the filtering by API queries. That way the
pagination count will be correct.
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
* Fix UI peers_from_control_nodes
Fixes bug where peers_from_control_node was
greyed out in UI.
Additional changes:
- Make edit instance button only show for instances
with managed = False
- Make remove instance button only show for instances
with managed = False
- InstanceList selectable only for instances with
managed = False
---------
Signed-off-by: Seth Foster <fosterbseth@gmail.com>
Works by adding a dedicated producer in wsrelay that looks for
local django channels message with group "metrics". The producer
sends this to the consumer running in the web container.
The consumer running in the web container handles the message by
pushing it into the local redis instance.
The django view that handles a request at the /api/v2/metrics
endpoint will load this data from redis, format it, and return the
response.
Seth FosterandPeter Braun
