Pin channels-redis to 4.3.1 to fix an async issue (#13348)

Refs django/channels_redis#332
Refs #13313

Signed-off-by: Rick Elrod <rick@elrod.me>

.github/triage_replies.md

@@ -53,6 +53,16 @@ https://github.com/ansible/awx/#get-involved \
 Thank you once again for this and your interest in AWX!
 
 
+### Red Hat Support Team
+- Hi! \
+\
+It appears that you are using an RPM build for RHEL. Please reach out to the Red Hat support team and submit a ticket. \
+\
+Here is the link to do so: \
+\
+https://access.redhat.com/support \
+\
+Thank you for your submission and for supporting AWX!
 
 
 ## Common

.github/workflows/ci.yml

@@ -2,6 +2,7 @@
 name: CI
 env:
   BRANCH: ${{ github.base_ref || 'devel' }}
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
 on:
   pull_request:
 jobs:
@@ -27,6 +28,9 @@ jobs:
           - name: awx-collection
             command: /start_tests.sh test_collection_all
             label: Run Collection Tests
+          - name: awx-collection-sanity
+            command: /start_tests.sh test_collection_sanity
+            label: Run Ansible core Collection Sanity tests
           - name: api-schema
             label: Check API Schema
             command: /start_tests.sh detect-schema-change SCHEMA_DIFF_BASE_BRANCH=${{ github.event.pull_request.base.ref }}

.github/workflows/devel_images.yml

@@ -1,5 +1,7 @@
 ---
 name: Build/Push Development Images
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
 on:
   push:
     branches:

.github/workflows/e2e_test.yml

@@ -1,5 +1,8 @@
 ---
 name: E2E Tests
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
+
 on:
   pull_request_target:
     types: [labeled]

.github/workflows/feature_branch_deletion.yml

@@ -1,5 +1,7 @@
 ---
 name: Feature branch deletion cleanup
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
 on:
   delete:
     branches:

.github/workflows/promote.yml

@@ -1,5 +1,9 @@
 ---
 name: Promote Release
+
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
+
 on:
   release:
     types: [published]

.github/workflows/stage.yml

@@ -1,5 +1,9 @@
 ---
 name: Stage Release
+
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
+
 on:
   workflow_dispatch:
     inputs:

.github/workflows/upload_schema.yml

@@ -1,5 +1,9 @@
 ---
 name: Upload API Schema
+
+env:
+  LC_ALL: "C.UTF-8" # prevent ERROR: Ansible could not initialize the preferred locale: unsupported locale setting
+
 on:
   push:
     branches:

MANIFEST.in

@@ -12,7 +12,7 @@ recursive-include awx/plugins *.ps1
 recursive-include requirements *.txt
 recursive-include requirements *.yml
 recursive-include config *
-recursive-include docs/licenses *
+recursive-include licenses *
 recursive-exclude awx devonly.py*
 recursive-exclude awx/api/tests *
 recursive-exclude awx/main/tests *

Makefile

@@ -6,7 +6,9 @@ CHROMIUM_BIN=/tmp/chrome-linux/chrome
 GIT_BRANCH ?= $(shell git rev-parse --abbrev-ref HEAD)
 MANAGEMENT_COMMAND ?= awx-manage
 VERSION := $(shell $(PYTHON) tools/scripts/scm_version.py)
-COLLECTION_VERSION := $(shell $(PYTHON) tools/scripts/scm_version.py | cut -d . -f 1-3)
+
+# ansible-test requires semver compatable version, so we allow overrides to hack it
+COLLECTION_VERSION ?= $(shell $(PYTHON) tools/scripts/scm_version.py | cut -d . -f 1-3)
 
 # NOTE: This defaults the container image version to the branch that's active
 COMPOSE_TAG ?= $(GIT_BRANCH)
@@ -34,7 +36,7 @@ RECEPTOR_IMAGE ?= quay.io/ansible/receptor:devel
 SRC_ONLY_PKGS ?= cffi,pycparser,psycopg2,twilio
 # These should be upgraded in the AWX and Ansible venv before attempting
 # to install the actual requirements
-VENV_BOOTSTRAP ?= pip==21.2.4 setuptools==58.2.0 setuptools_scm[toml]==6.4.2 wheel==0.36.2
+VENV_BOOTSTRAP ?= pip==21.2.4 setuptools==65.6.3 setuptools_scm[toml]==7.0.5 wheel==0.38.4
 
 NAME ?= awx
 
@@ -118,7 +120,7 @@ virtualenv_awx:
 		fi; \
 	fi
 
-## Install third-party requirements needed for AWX's environment. 
+## Install third-party requirements needed for AWX's environment.
 # this does not use system site packages intentionally
 requirements_awx: virtualenv_awx
 	if [[ "$(PIP_OPTIONS)" == *"--no-index"* ]]; then \
@@ -300,7 +302,8 @@ test_collection:
 	if [ "$(VENV_BASE)" ]; then \
 		. $(VENV_BASE)/awx/bin/activate; \
 	fi && \
-	pip install ansible-core && \
+	if ! [ -x "$(shell command -v ansible-playbook)" ]; then pip install ansible-core; fi
+	ansible --version
 	py.test $(COLLECTION_TEST_DIRS) -v
 	# The python path needs to be modified so that the tests can find Ansible within the container
 	# First we will use anything expility set as PYTHONPATH
@@ -330,8 +333,13 @@ install_collection: build_collection
 	rm -rf $(COLLECTION_INSTALL)
 	ansible-galaxy collection install awx_collection_build/$(COLLECTION_NAMESPACE)-$(COLLECTION_PACKAGE)-$(COLLECTION_VERSION).tar.gz
 
-test_collection_sanity: install_collection
-	cd $(COLLECTION_INSTALL) && ansible-test sanity
+test_collection_sanity:
+	rm -rf awx_collection_build/
+	rm -rf $(COLLECTION_INSTALL)
+	if ! [ -x "$(shell command -v ansible-test)" ]; then pip install ansible-core; fi
+	ansible --version
+	COLLECTION_VERSION=1.0.0 make install_collection
+	cd $(COLLECTION_INSTALL) && ansible-test sanity --exclude=plugins/modules/export.py
 
 test_collection_integration: install_collection
 	cd $(COLLECTION_INSTALL) && ansible-test integration $(COLLECTION_TEST_TARGET)
@@ -389,18 +397,18 @@ $(UI_BUILD_FLAG_FILE):
 	$(PYTHON) tools/scripts/compilemessages.py
 	$(NPM_BIN) --prefix awx/ui --loglevel warn run compile-strings
 	$(NPM_BIN) --prefix awx/ui --loglevel warn run build
-	mkdir -p /var/lib/awx/public/static/css
-	mkdir -p /var/lib/awx/public/static/js
-	mkdir -p /var/lib/awx/public/static/media
-	cp -r awx/ui/build/static/css/* /var/lib/awx/public/static/css
-	cp -r awx/ui/build/static/js/* /var/lib/awx/public/static/js
-	cp -r awx/ui/build/static/media/* /var/lib/awx/public/static/media
 	touch $@
 
 ui-release: $(UI_BUILD_FLAG_FILE)
 
 ui-devel: awx/ui/node_modules
 	@$(MAKE) -B $(UI_BUILD_FLAG_FILE)
+	mkdir -p /var/lib/awx/public/static/css
+	mkdir -p /var/lib/awx/public/static/js
+	mkdir -p /var/lib/awx/public/static/media
+	cp -r awx/ui/build/static/css/* /var/lib/awx/public/static/css
+	cp -r awx/ui/build/static/js/* /var/lib/awx/public/static/js
+	cp -r awx/ui/build/static/media/* /var/lib/awx/public/static/media
 
 ui-devel-instrumented: awx/ui/node_modules
 	$(NPM_BIN) --prefix awx/ui --loglevel warn run start-instrumented
@@ -452,7 +460,7 @@ awx/projects:
 COMPOSE_UP_OPTS ?=
 COMPOSE_OPTS ?=
 CONTROL_PLANE_NODE_COUNT ?= 1
-EXECUTION_NODE_COUNT ?= 2
+EXECUTION_NODE_COUNT ?= 0
 MINIKUBE_CONTAINER_GROUP ?= false
 MINIKUBE_SETUP ?= false # if false, run minikube separately
 EXTRA_SOURCES_ANSIBLE_OPTS ?=
@@ -593,13 +601,12 @@ pot: $(UI_BUILD_FLAG_FILE)
 po: $(UI_BUILD_FLAG_FILE)
 	$(NPM_BIN) --prefix awx/ui --loglevel warn run extract-strings -- --clean
 
-LANG = "en_us"
 ## generate API django .pot .po
 messages:
 	@if [ "$(VENV_BASE)" ]; then \
 		. $(VENV_BASE)/awx/bin/activate; \
 	fi; \
-	$(PYTHON) manage.py makemessages -l $(LANG) --keep-pot
+	$(PYTHON) manage.py makemessages -l en_us --keep-pot
 
 print-%:
 	@echo $($*)

awx/api/serializers.py

@@ -113,7 +113,7 @@ from awx.main.utils import (
 )
 from awx.main.utils.filters import SmartFilter
 from awx.main.utils.named_url_graph import reset_counters
-from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups, TaskManagerInstances
+from awx.main.scheduler.task_manager_models import TaskManagerModels
 from awx.main.redact import UriCleaner, REPLACE_STR
 
 from awx.main.validators import vars_validate_or_raise
@@ -5040,12 +5040,10 @@ class InstanceHealthCheckSerializer(BaseSerializer):
 class InstanceGroupSerializer(BaseSerializer):
 
     show_capabilities = ['edit', 'delete']
-
+    capacity = serializers.SerializerMethodField()
     consumed_capacity = serializers.SerializerMethodField()
     percent_capacity_remaining = serializers.SerializerMethodField()
-    jobs_running = serializers.IntegerField(
-        help_text=_('Count of jobs in the running or waiting state that ' 'are targeted for this instance group'), read_only=True
-    )
+    jobs_running = serializers.SerializerMethodField()
     jobs_total = serializers.IntegerField(help_text=_('Count of all jobs that target this instance group'), read_only=True)
     instances = serializers.SerializerMethodField()
     is_container_group = serializers.BooleanField(
@@ -5071,6 +5069,22 @@ class InstanceGroupSerializer(BaseSerializer):
         label=_('Policy Instance Minimum'),
         help_text=_("Static minimum number of Instances that will be automatically assign to " "this group when new instances come online."),
     )
+    max_concurrent_jobs = serializers.IntegerField(
+        default=0,
+        min_value=0,
+        required=False,
+        initial=0,
+        label=_('Max Concurrent Jobs'),
+        help_text=_("Maximum number of concurrent jobs to run on a group. When set to zero, no maximum is enforced."),
+    )
+    max_forks = serializers.IntegerField(
+        default=0,
+        min_value=0,
+        required=False,
+        initial=0,
+        label=_('Max Forks'),
+        help_text=_("Maximum number of forks to execute concurrently on a group. When set to zero, no maximum is enforced."),
+    )
     policy_instance_list = serializers.ListField(
         child=serializers.CharField(),
         required=False,
@@ -5092,6 +5106,8 @@ class InstanceGroupSerializer(BaseSerializer):
             "consumed_capacity",
             "percent_capacity_remaining",
             "jobs_running",
+            "max_concurrent_jobs",
+            "max_forks",
             "jobs_total",
             "instances",
             "is_container_group",
@@ -5173,28 +5189,39 @@ class InstanceGroupSerializer(BaseSerializer):
         # Store capacity values (globally computed) in the context
         if 'task_manager_igs' not in self.context:
             instance_groups_queryset = None
-            jobs_qs = UnifiedJob.objects.filter(status__in=('running', 'waiting'))
             if self.parent:  # Is ListView:
                 instance_groups_queryset = self.parent.instance
 
-            instances = TaskManagerInstances(jobs_qs)
-            instance_groups = TaskManagerInstanceGroups(instances_by_hostname=instances, instance_groups_queryset=instance_groups_queryset)
+            tm_models = TaskManagerModels.init_with_consumed_capacity(
+                instance_fields=['uuid', 'version', 'capacity', 'cpu', 'memory', 'managed_by_policy', 'enabled'],
+                instance_groups_queryset=instance_groups_queryset,
+            )
 
-            self.context['task_manager_igs'] = instance_groups
+            self.context['task_manager_igs'] = tm_models.instance_groups
         return self.context['task_manager_igs']
 
     def get_consumed_capacity(self, obj):
         ig_mgr = self.get_ig_mgr()
         return ig_mgr.get_consumed_capacity(obj.name)
 
-    def get_percent_capacity_remaining(self, obj):
-        if not obj.capacity:
-            return 0.0
+    def get_capacity(self, obj):
         ig_mgr = self.get_ig_mgr()
-        return float("{0:.2f}".format((float(ig_mgr.get_remaining_capacity(obj.name)) / (float(obj.capacity))) * 100))
+        return ig_mgr.get_capacity(obj.name)
+
+    def get_percent_capacity_remaining(self, obj):
+        capacity = self.get_capacity(obj)
+        if not capacity:
+            return 0.0
+        consumed_capacity = self.get_consumed_capacity(obj)
+        return float("{0:.2f}".format(((float(capacity) - float(consumed_capacity)) / (float(capacity))) * 100))
 
     def get_instances(self, obj):
-        return obj.instances.count()
+        ig_mgr = self.get_ig_mgr()
+        return len(ig_mgr.get_instances(obj.name))
+
+    def get_jobs_running(self, obj):
+        ig_mgr = self.get_ig_mgr()
+        return ig_mgr.get_jobs_running(obj.name)
 
 
 class ActivityStreamSerializer(BaseSerializer):

awx/api/views/__init__.py

@@ -5,6 +5,7 @@
 import dateutil
 import functools
 import html
+import itertools
 import logging
 import re
 import requests
@@ -20,9 +21,10 @@ from urllib3.exceptions import ConnectTimeoutError
 # Django
 from django.conf import settings
 from django.core.exceptions import FieldError, ObjectDoesNotExist
-from django.db.models import Q, Sum
+from django.db.models import Q, Sum, Count
 from django.db import IntegrityError, ProgrammingError, transaction, connection
 from django.db.models.fields.related import ManyToManyField, ForeignKey
+from django.db.models.functions import Trunc
 from django.shortcuts import get_object_or_404
 from django.utils.safestring import mark_safe
 from django.utils.timezone import now
@@ -47,9 +49,6 @@ from rest_framework import status
 from rest_framework_yaml.parsers import YAMLParser
 from rest_framework_yaml.renderers import YAMLRenderer
 
-# QSStats
-import qsstats
-
 # ANSIConv
 import ansiconv
 
@@ -283,30 +282,50 @@ class DashboardJobsGraphView(APIView):
             success_query = success_query.filter(instance_of=models.ProjectUpdate)
             failed_query = failed_query.filter(instance_of=models.ProjectUpdate)
 
-        success_qss = qsstats.QuerySetStats(success_query, 'finished')
-        failed_qss = qsstats.QuerySetStats(failed_query, 'finished')
-
-        start_date = now()
+        end = now()
+        interval = 'day'
         if period == 'month':
-            end_date = start_date - dateutil.relativedelta.relativedelta(months=1)
-            interval = 'days'
+            start = end - dateutil.relativedelta.relativedelta(months=1)
         elif period == 'two_weeks':
-            end_date = start_date - dateutil.relativedelta.relativedelta(weeks=2)
-            interval = 'days'
+            start = end - dateutil.relativedelta.relativedelta(weeks=2)
         elif period == 'week':
-            end_date = start_date - dateutil.relativedelta.relativedelta(weeks=1)
-            interval = 'days'
+            start = end - dateutil.relativedelta.relativedelta(weeks=1)
         elif period == 'day':
-            end_date = start_date - dateutil.relativedelta.relativedelta(days=1)
-            interval = 'hours'
+            start = end - dateutil.relativedelta.relativedelta(days=1)
+            interval = 'hour'
         else:
             return Response({'error': _('Unknown period "%s"') % str(period)}, status=status.HTTP_400_BAD_REQUEST)
 
         dashboard_data = {"jobs": {"successful": [], "failed": []}}
-        for element in success_qss.time_series(end_date, start_date, interval=interval):
-            dashboard_data['jobs']['successful'].append([time.mktime(element[0].timetuple()), element[1]])
-        for element in failed_qss.time_series(end_date, start_date, interval=interval):
-            dashboard_data['jobs']['failed'].append([time.mktime(element[0].timetuple()), element[1]])
+
+        succ_list = dashboard_data['jobs']['successful']
+        fail_list = dashboard_data['jobs']['failed']
+
+        qs_s = (
+            success_query.filter(finished__range=(start, end))
+            .annotate(d=Trunc('finished', interval, tzinfo=end.tzinfo))
+            .order_by()
+            .values('d')
+            .annotate(agg=Count('id', distinct=True))
+        )
+        data_s = {item['d']: item['agg'] for item in qs_s}
+        qs_f = (
+            failed_query.filter(finished__range=(start, end))
+            .annotate(d=Trunc('finished', interval, tzinfo=end.tzinfo))
+            .order_by()
+            .values('d')
+            .annotate(agg=Count('id', distinct=True))
+        )
+        data_f = {item['d']: item['agg'] for item in qs_f}
+
+        start_date = start.replace(hour=0, minute=0, second=0, microsecond=0)
+        for d in itertools.count():
+            date = start_date + dateutil.relativedelta.relativedelta(days=d)
+            if date > end:
+                break
+            succ_list.append([time.mktime(date.timetuple()), data_s.get(date, 0)])
+            fail_list.append([time.mktime(date.timetuple()), data_f.get(date, 0)])
+
         return Response(dashboard_data)
 
 
@@ -325,6 +344,13 @@ class InstanceDetail(RetrieveUpdateAPIView):
     model = models.Instance
     serializer_class = serializers.InstanceSerializer
 
+    def update_raw_data(self, data):
+        # these fields are only valid on creation of an instance, so they unwanted on detail view
+        data.pop('listener_port', None)
+        data.pop('node_type', None)
+        data.pop('hostname', None)
+        return super(InstanceDetail, self).update_raw_data(data)
+
     def update(self, request, *args, **kwargs):
         r = super(InstanceDetail, self).update(request, *args, **kwargs)
         if status.is_success(r.status_code):

awx/api/views/mixin.py

@@ -16,7 +16,7 @@ from rest_framework import status
 
 from awx.main.constants import ACTIVE_STATES
 from awx.main.utils import get_object_or_400
-from awx.main.models.ha import Instance, InstanceGroup
+from awx.main.models.ha import Instance, InstanceGroup, schedule_policy_task
 from awx.main.models.organization import Team
 from awx.main.models.projects import Project
 from awx.main.models.inventory import Inventory
@@ -107,6 +107,11 @@ class InstanceGroupMembershipMixin(object):
                 if inst_name in ig_obj.policy_instance_list:
                     ig_obj.policy_instance_list.pop(ig_obj.policy_instance_list.index(inst_name))
                     ig_obj.save(update_fields=['policy_instance_list'])
+
+            # sometimes removing an instance has a non-obvious consequence
+            # this is almost always true if policy_instance_percentage or _minimum is non-zero
+            # after removing a single instance, the other memberships need to be re-balanced
+            schedule_policy_task()
         return response
 
 

awx/locale/es/LC_MESSAGES/django.po

@@ -6237,4 +6237,5 @@ msgstr "%s se está actualizando."
 
 #: awx/ui/urls.py:24
 msgid "This page will refresh when complete."
-msgstr "Esta página se actualizará cuando se complete."
+msgstr "Esta página se actualizará cuando se complete."
+

awx/locale/fr/LC_MESSAGES/django.po

@@ -721,7 +721,7 @@ msgstr "DTSTART valide obligatoire dans rrule. La valeur doit commencer par : DT
 #: awx/api/serializers.py:4657
 msgid ""
 "DTSTART cannot be a naive datetime.  Specify ;TZINFO= or YYYYMMDDTHHMMSSZZ."
-msgstr "DTSTART ne peut correspondre à une DateHeure naïve. Spécifier ;TZINFO= ou YYYYMMDDTHHMMSSZZ."
+msgstr "DTSTART ne peut correspondre à une date-heure naïve. Spécifier ;TZINFO= ou YYYYMMDDTHHMMSSZZ."
 
 #: awx/api/serializers.py:4659
 msgid "Multiple DTSTART is not supported."
@@ -6239,4 +6239,5 @@ msgstr "%s est en cours de mise à niveau."
 
 #: awx/ui/urls.py:24
 msgid "This page will refresh when complete."
-msgstr "Cette page sera rafraîchie une fois terminée."
+msgstr "Cette page sera rafraîchie une fois terminée."
+

awx/locale/nl/LC_MESSAGES/django.po

@@ -6237,4 +6237,5 @@ msgstr "Er wordt momenteel een upgrade van%s geïnstalleerd."
 
 #: awx/ui/urls.py:24
 msgid "This page will refresh when complete."
-msgstr "Deze pagina wordt vernieuwd als hij klaar is."
+msgstr "Deze pagina wordt vernieuwd als hij klaar is."
+

awx/main/access.py

@@ -2697,46 +2697,66 @@ class ActivityStreamAccess(BaseAccess):
         # 'job_template', 'job', 'project', 'project_update', 'workflow_job',
         # 'inventory_source', 'workflow_job_template'
 
-        inventory_set = Inventory.accessible_objects(self.user, 'read_role')
-        credential_set = Credential.accessible_objects(self.user, 'read_role')
+        q = Q(user=self.user)
+        inventory_set = Inventory.accessible_pk_qs(self.user, 'read_role')
+        if inventory_set:
+            q |= (
+                Q(ad_hoc_command__inventory__in=inventory_set)
+                | Q(inventory__in=inventory_set)
+                | Q(host__inventory__in=inventory_set)
+                | Q(group__inventory__in=inventory_set)
+                | Q(inventory_source__inventory__in=inventory_set)
+                | Q(inventory_update__inventory_source__inventory__in=inventory_set)
+            )
+
+        credential_set = Credential.accessible_pk_qs(self.user, 'read_role')
+        if credential_set:
+            q |= Q(credential__in=credential_set)
+
         auditing_orgs = (
             (Organization.accessible_objects(self.user, 'admin_role') | Organization.accessible_objects(self.user, 'auditor_role'))
             .distinct()
             .values_list('id', flat=True)
         )
-        project_set = Project.accessible_objects(self.user, 'read_role')
-        jt_set = JobTemplate.accessible_objects(self.user, 'read_role')
-        team_set = Team.accessible_objects(self.user, 'read_role')
-        wfjt_set = WorkflowJobTemplate.accessible_objects(self.user, 'read_role')
-        app_set = OAuth2ApplicationAccess(self.user).filtered_queryset()
-        token_set = OAuth2TokenAccess(self.user).filtered_queryset()
+        if auditing_orgs:
+            q |= (
+                Q(user__in=auditing_orgs.values('member_role__members'))
+                | Q(organization__in=auditing_orgs)
+                | Q(notification_template__organization__in=auditing_orgs)
+                | Q(notification__notification_template__organization__in=auditing_orgs)
+                | Q(label__organization__in=auditing_orgs)
+                | Q(role__in=Role.objects.filter(ancestors__in=self.user.roles.all()) if auditing_orgs else [])
+            )
 
-        return qs.filter(
-            Q(ad_hoc_command__inventory__in=inventory_set)
-            | Q(o_auth2_application__in=app_set)
-            | Q(o_auth2_access_token__in=token_set)
-            | Q(user__in=auditing_orgs.values('member_role__members'))
-            | Q(user=self.user)
-            | Q(organization__in=auditing_orgs)
-            | Q(inventory__in=inventory_set)
-            | Q(host__inventory__in=inventory_set)
-            | Q(group__inventory__in=inventory_set)
-            | Q(inventory_source__inventory__in=inventory_set)
-            | Q(inventory_update__inventory_source__inventory__in=inventory_set)
-            | Q(credential__in=credential_set)
-            | Q(team__in=team_set)
-            | Q(project__in=project_set)
-            | Q(project_update__project__in=project_set)
-            | Q(job_template__in=jt_set)
-            | Q(job__job_template__in=jt_set)
-            | Q(workflow_job_template__in=wfjt_set)
-            | Q(workflow_job_template_node__workflow_job_template__in=wfjt_set)
-            | Q(workflow_job__workflow_job_template__in=wfjt_set)
-            | Q(notification_template__organization__in=auditing_orgs)
-            | Q(notification__notification_template__organization__in=auditing_orgs)
-            | Q(label__organization__in=auditing_orgs)
-            | Q(role__in=Role.objects.filter(ancestors__in=self.user.roles.all()) if auditing_orgs else [])
-        ).distinct()
+        project_set = Project.accessible_pk_qs(self.user, 'read_role')
+        if project_set:
+            q |= Q(project__in=project_set) | Q(project_update__project__in=project_set)
+
+        jt_set = JobTemplate.accessible_pk_qs(self.user, 'read_role')
+        if jt_set:
+            q |= Q(job_template__in=jt_set) | Q(job__job_template__in=jt_set)
+
+        wfjt_set = WorkflowJobTemplate.accessible_pk_qs(self.user, 'read_role')
+        if wfjt_set:
+            q |= (
+                Q(workflow_job_template__in=wfjt_set)
+                | Q(workflow_job_template_node__workflow_job_template__in=wfjt_set)
+                | Q(workflow_job__workflow_job_template__in=wfjt_set)
+            )
+
+        team_set = Team.accessible_pk_qs(self.user, 'read_role')
+        if team_set:
+            q |= Q(team__in=team_set)
+
+        app_set = OAuth2ApplicationAccess(self.user).filtered_queryset()
+        if app_set:
+            q |= Q(o_auth2_application__in=app_set)
+
+        token_set = OAuth2TokenAccess(self.user).filtered_queryset()
+        if token_set:
+            q |= Q(o_auth2_access_token__in=token_set)
+
+        return qs.filter(q).distinct()
 
     def can_add(self, data):
         return False

awx/main/analytics/broadcast_websocket.py

@@ -1,8 +1,8 @@
 import datetime
 import asyncio
 import logging
-import aioredis
 import redis
+import redis.asyncio
 import re
 
 from prometheus_client import (
@@ -82,7 +82,7 @@ class BroadcastWebsocketStatsManager:
 
     async def run_loop(self):
         try:
-            redis_conn = await aioredis.create_redis_pool(settings.BROKER_URL)
+            redis_conn = await redis.asyncio.Redis.from_url(settings.BROKER_URL)
             while True:
                 stats_data_str = ''.join(stat.serialize() for stat in self._stats.values())
                 await redis_conn.set(self._redis_key, stats_data_str)
@@ -122,8 +122,8 @@ class BroadcastWebsocketStats:
             'Number of messages received, to be forwarded, by the broadcast websocket system',
             registry=self._registry,
         )
-        self._messages_received = Gauge(
-            f'awx_{self.remote_name}_messages_received',
+        self._messages_received_current_conn = Gauge(
+            f'awx_{self.remote_name}_messages_received_currrent_conn',
             'Number forwarded messages received by the broadcast websocket system, for the duration of the current connection',
             registry=self._registry,
         )
@@ -144,13 +144,13 @@ class BroadcastWebsocketStats:
 
     def record_message_received(self):
         self._internal_messages_received_per_minute.record()
-        self._messages_received.inc()
+        self._messages_received_current_conn.inc()
         self._messages_received_total.inc()
 
     def record_connection_established(self):
         self._connection.state('connected')
         self._connection_start.set_to_current_time()
-        self._messages_received.set(0)
+        self._messages_received_current_conn.set(0)
 
     def record_connection_lost(self):
         self._connection.state('disconnected')

awx/main/analytics/collectors.py

@@ -16,7 +16,7 @@ from awx.conf.license import get_license
 from awx.main.utils import get_awx_version, camelcase_to_underscore, datetime_hook
 from awx.main import models
 from awx.main.analytics import register
-from awx.main.scheduler.task_manager_models import TaskManagerInstances
+from awx.main.scheduler.task_manager_models import TaskManagerModels
 
 """
 This module is used to define metrics collected by awx.main.analytics.gather()
@@ -237,9 +237,8 @@ def projects_by_scm_type(since, **kwargs):
 def instance_info(since, include_hostnames=False, **kwargs):
     info = {}
     # Use same method that the TaskManager does to compute consumed capacity without querying all running jobs for each Instance
-    active_tasks = models.UnifiedJob.objects.filter(status__in=['running', 'waiting']).only('task_impact', 'controller_node', 'execution_node')
-    tm_instances = TaskManagerInstances(active_tasks, instance_fields=['uuid', 'version', 'capacity', 'cpu', 'memory', 'managed_by_policy', 'enabled'])
-    for tm_instance in tm_instances.instances_by_hostname.values():
+    tm_models = TaskManagerModels.init_with_consumed_capacity(instance_fields=['uuid', 'version', 'capacity', 'cpu', 'memory', 'managed_by_policy', 'enabled'])
+    for tm_instance in tm_models.instances.instances_by_hostname.values():
         instance = tm_instance.obj
         instance_info = {
             'uuid': instance.uuid,
@@ -251,6 +250,7 @@ def instance_info(since, include_hostnames=False, **kwargs):
             'enabled': instance.enabled,
             'consumed_capacity': tm_instance.consumed_capacity,
             'remaining_capacity': instance.capacity - tm_instance.consumed_capacity,
+            'node_type': instance.node_type,
         }
         if include_hostnames is True:
             instance_info['hostname'] = instance.hostname

awx/main/analytics/metrics.py

@@ -57,6 +57,7 @@ def metrics():
         [
             'hostname',
             'instance_uuid',
+            'node_type',
         ],
         registry=REGISTRY,
     )
@@ -84,6 +85,7 @@ def metrics():
         [
             'hostname',
             'instance_uuid',
+            'node_type',
         ],
         registry=REGISTRY,
     )
@@ -111,6 +113,7 @@ def metrics():
         [
             'hostname',
             'instance_uuid',
+            'node_type',
         ],
         registry=REGISTRY,
     )
@@ -120,6 +123,7 @@ def metrics():
         [
             'hostname',
             'instance_uuid',
+            'node_type',
         ],
         registry=REGISTRY,
     )
@@ -180,12 +184,13 @@ def metrics():
     instance_data = instance_info(None, include_hostnames=True)
     for uuid, info in instance_data.items():
         hostname = info['hostname']
-        INSTANCE_CAPACITY.labels(hostname=hostname, instance_uuid=uuid).set(instance_data[uuid]['capacity'])
+        node_type = info['node_type']
+        INSTANCE_CAPACITY.labels(hostname=hostname, instance_uuid=uuid, node_type=node_type).set(instance_data[uuid]['capacity'])
         INSTANCE_CPU.labels(hostname=hostname, instance_uuid=uuid).set(instance_data[uuid]['cpu'])
         INSTANCE_MEMORY.labels(hostname=hostname, instance_uuid=uuid).set(instance_data[uuid]['memory'])
-        INSTANCE_CONSUMED_CAPACITY.labels(hostname=hostname, instance_uuid=uuid).set(instance_data[uuid]['consumed_capacity'])
-        INSTANCE_REMAINING_CAPACITY.labels(hostname=hostname, instance_uuid=uuid).set(instance_data[uuid]['remaining_capacity'])
-        INSTANCE_INFO.labels(hostname=hostname, instance_uuid=uuid).info(
+        INSTANCE_CONSUMED_CAPACITY.labels(hostname=hostname, instance_uuid=uuid, node_type=node_type).set(instance_data[uuid]['consumed_capacity'])
+        INSTANCE_REMAINING_CAPACITY.labels(hostname=hostname, instance_uuid=uuid, node_type=node_type).set(instance_data[uuid]['remaining_capacity'])
+        INSTANCE_INFO.labels(hostname=hostname, instance_uuid=uuid, node_type=node_type).info(
             {
                 'enabled': str(instance_data[uuid]['enabled']),
                 'managed_by_policy': str(instance_data[uuid]['managed_by_policy']),

awx/main/analytics/subsystem_metrics.py

@@ -5,7 +5,9 @@ import logging
 
 from django.conf import settings
 from django.apps import apps
+
 from awx.main.consumers import emit_channel_notification
+from awx.main.utils import is_testing
 
 root_key = 'awx_metrics'
 logger = logging.getLogger('awx.main.analytics')
@@ -163,7 +165,7 @@ class Metrics:
         Instance = apps.get_model('main', 'Instance')
         if instance_name:
             self.instance_name = instance_name
-        elif settings.IS_TESTING():
+        elif is_testing():
             self.instance_name = "awx_testing"
         else:
             self.instance_name = Instance.objects.my_hostname()

awx/main/conf.py

@@ -569,7 +569,7 @@ register(
 register(
     'LOG_AGGREGATOR_LOGGERS',
     field_class=fields.StringListField,
-    default=['awx', 'activity_stream', 'job_events', 'system_tracking'],
+    default=['awx', 'activity_stream', 'job_events', 'system_tracking', 'broadcast_websocket'],
     label=_('Loggers Sending Data to Log Aggregator Form'),
     help_text=_(
         'List of loggers that will send HTTP logs to the collector, these can '
@@ -577,7 +577,8 @@ register(
         'awx - service logs\n'
         'activity_stream - activity stream records\n'
         'job_events - callback data from Ansible job events\n'
-        'system_tracking - facts gathered from scan jobs.'
+        'system_tracking - facts gathered from scan jobs\n'
+        'broadcast_websocket - errors pertaining to websockets broadcast metrics\n'
     ),
     category=_('Logging'),
     category_slug='logging',

awx/main/dispatch/pool.py

@@ -466,7 +466,7 @@ class AutoscalePool(WorkerPool):
                 task_name = 'unknown'
                 if isinstance(body, dict):
                     task_name = body.get('task')
-                logger.warn(f'Workers maxed, queuing {task_name}, load: {sum(len(w.managed_tasks) for w in self.workers)} / {len(self.workers)}')
+                logger.warning(f'Workers maxed, queuing {task_name}, load: {sum(len(w.managed_tasks) for w in self.workers)} / {len(self.workers)}')
                 return super(AutoscalePool, self).write(preferred_queue, body)
         except Exception:
             for conn in connections.all():

awx/main/dispatch/publish.py

@@ -1,14 +1,13 @@
 import inspect
 import logging
-import sys
 import json
 import time
 from uuid import uuid4
 
-from django.conf import settings
 from django_guid import get_guid
 
 from . import pg_bus_conn
+from awx.main.utils import is_testing
 
 logger = logging.getLogger('awx.main.dispatch')
 
@@ -93,7 +92,7 @@ class task:
                 obj.update(**kw)
                 if callable(queue):
                     queue = queue()
-                if not settings.IS_TESTING(sys.argv):
+                if not is_testing():
                     with pg_bus_conn() as conn:
                         conn.notify(queue, json.dumps(obj))
                 return (obj, queue)

awx/main/management/commands/provision_instance.py

@@ -38,7 +38,14 @@ class Command(BaseCommand):
             (changed, instance) = Instance.objects.register(ip_address=os.environ.get('MY_POD_IP'), node_type='control', uuid=settings.SYSTEM_UUID)
             RegisterQueue(settings.DEFAULT_CONTROL_PLANE_QUEUE_NAME, 100, 0, [], is_container_group=False).register()
             RegisterQueue(
-                settings.DEFAULT_EXECUTION_QUEUE_NAME, 100, 0, [], is_container_group=True, pod_spec_override=settings.DEFAULT_EXECUTION_QUEUE_POD_SPEC_OVERRIDE
+                settings.DEFAULT_EXECUTION_QUEUE_NAME,
+                100,
+                0,
+                [],
+                is_container_group=True,
+                pod_spec_override=settings.DEFAULT_EXECUTION_QUEUE_POD_SPEC_OVERRIDE,
+                max_forks=settings.DEFAULT_EXECUTION_QUEUE_MAX_FORKS,
+                max_concurrent_jobs=settings.DEFAULT_EXECUTION_QUEUE_MAX_CONCURRENT_JOBS,
             ).register()
         else:
             (changed, instance) = Instance.objects.register(hostname=hostname, node_type=node_type, uuid=uuid)

awx/main/management/commands/regenerate_secret_key.py

@@ -32,8 +32,14 @@ class Command(BaseCommand):
     def handle(self, **options):
         self.old_key = settings.SECRET_KEY
         custom_key = os.environ.get("TOWER_SECRET_KEY")
-        if options.get("use_custom_key") and custom_key:
-            self.new_key = custom_key
+        if options.get("use_custom_key"):
+            if custom_key:
+                self.new_key = custom_key
+            else:
+                print("Use custom key was specified but the env var TOWER_SECRET_KEY was not available")
+                import sys
+
+                sys.exit(1)
         else:
             self.new_key = base64.encodebytes(os.urandom(33)).decode().rstrip()
         self._notification_templates()

awx/main/management/commands/register_queue.py

@@ -17,7 +17,9 @@ class InstanceNotFound(Exception):
 
 
 class RegisterQueue:
-    def __init__(self, queuename, instance_percent, inst_min, hostname_list, is_container_group=None, pod_spec_override=None):
+    def __init__(
+        self, queuename, instance_percent, inst_min, hostname_list, is_container_group=None, pod_spec_override=None, max_forks=None, max_concurrent_jobs=None
+    ):
         self.instance_not_found_err = None
         self.queuename = queuename
         self.instance_percent = instance_percent
@@ -25,6 +27,8 @@ class RegisterQueue:
         self.hostname_list = hostname_list
         self.is_container_group = is_container_group
         self.pod_spec_override = pod_spec_override
+        self.max_forks = max_forks
+        self.max_concurrent_jobs = max_concurrent_jobs
 
     def get_create_update_instance_group(self):
         created = False
@@ -45,6 +49,14 @@ class RegisterQueue:
             ig.pod_spec_override = self.pod_spec_override
             changed = True
 
+        if self.max_forks and (ig.max_forks != self.max_forks):
+            ig.max_forks = self.max_forks
+            changed = True
+
+        if self.max_concurrent_jobs and (ig.max_concurrent_jobs != self.max_concurrent_jobs):
+            ig.max_concurrent_jobs = self.max_concurrent_jobs
+            changed = True
+
         if changed:
             ig.save()
 

awx/main/managers.py

@@ -158,7 +158,11 @@ class InstanceManager(models.Manager):
                     return (False, instance)
 
             # Create new instance, and fill in default values
-            create_defaults = {'node_state': Instance.States.INSTALLED, 'capacity': 0}
+            create_defaults = {
+                'node_state': Instance.States.INSTALLED,
+                'capacity': 0,
+                'listener_port': 27199,
+            }
             if defaults is not None:
                 create_defaults.update(defaults)
             uuid_option = {}

awx/main/migrations/0164_remove_inventorysource_update_on_project_update.py

@@ -1,24 +1,14 @@
 # Generated by Django 3.2.13 on 2022-06-21 21:29
 
 from django.db import migrations
-import logging
-
-logger = logging.getLogger("awx")
 
 
 def forwards(apps, schema_editor):
     InventorySource = apps.get_model('main', 'InventorySource')
-    sources = InventorySource.objects.filter(update_on_project_update=True)
-    for src in sources:
-        if src.update_on_launch == False:
-            src.update_on_launch = True
-            src.save(update_fields=['update_on_launch'])
-            logger.info(f"Setting update_on_launch to True for {src}")
-        proj = src.source_project
-        if proj and proj.scm_update_on_launch is False:
-            proj.scm_update_on_launch = True
-            proj.save(update_fields=['scm_update_on_launch'])
-            logger.warning(f"Setting scm_update_on_launch to True for {proj}")
+    InventorySource.objects.filter(update_on_project_update=True).update(update_on_launch=True)
+
+    Project = apps.get_model('main', 'Project')
+    Project.objects.filter(scm_inventory_sources__update_on_project_update=True).update(scm_update_on_launch=True)
 
 
 class Migration(migrations.Migration):

awx/main/migrations/0173_instancegroup_max_limits.py

@@ -0,0 +1,23 @@
+# Generated by Django 3.2.13 on 2022-10-24 18:22
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('main', '0172_prevent_instance_fallback'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='instancegroup',
+            name='max_concurrent_jobs',
+            field=models.IntegerField(default=0, help_text='Maximum number of concurrent jobs to run on this group. Zero means no limit.'),
+        ),
+        migrations.AddField(
+            model_name='instancegroup',
+            name='max_forks',
+            field=models.IntegerField(default=0, help_text='Max forks to execute on this group. Zero means no limit.'),
+        ),
+    ]

awx/main/migrations/0174_ensure_org_ee_admin_roles.py

@@ -0,0 +1,18 @@
+# Generated by Django 3.2.16 on 2022-12-07 21:11
+
+from django.db import migrations
+
+from awx.main.migrations import _rbac as rbac
+from awx.main.migrations import _migration_utils as migration_utils
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('main', '0173_instancegroup_max_limits'),
+    ]
+
+    operations = [
+        migrations.RunPython(migration_utils.set_current_apps_for_migrations),
+        migrations.RunPython(rbac.create_roles),
+    ]

awx/main/models/credential/injectors.py

@@ -15,6 +15,7 @@ def aws(cred, env, private_data_dir):
 
     if cred.has_input('security_token'):
         env['AWS_SECURITY_TOKEN'] = cred.get_input('security_token', default='')
+        env['AWS_SESSION_TOKEN'] = env['AWS_SECURITY_TOKEN']
 
 
 def gce(cred, env, private_data_dir):

awx/main/models/ha.py

@@ -233,11 +233,12 @@ class Instance(HasPolicyEditsMixin, BaseModel):
         if not isinstance(vargs.get('grace_period'), int):
             vargs['grace_period'] = 60  # grace period of 60 minutes, need to set because CLI default will not take effect
         if 'exclude_strings' not in vargs and vargs.get('file_pattern'):
-            active_pks = list(
-                UnifiedJob.objects.filter(
-                    (models.Q(execution_node=self.hostname) | models.Q(controller_node=self.hostname)) & models.Q(status__in=('running', 'waiting'))
-                ).values_list('pk', flat=True)
-            )
+            active_job_qs = UnifiedJob.objects.filter(status__in=('running', 'waiting'))
+            if self.node_type == 'execution':
+                active_job_qs = active_job_qs.filter(execution_node=self.hostname)
+            else:
+                active_job_qs = active_job_qs.filter(controller_node=self.hostname)
+            active_pks = list(active_job_qs.values_list('pk', flat=True))
             if active_pks:
                 vargs['exclude_strings'] = [JOB_FOLDER_PREFIX % job_id for job_id in active_pks]
         if 'remove_images' in vargs or 'image_prune' in vargs:
@@ -378,6 +379,8 @@ class InstanceGroup(HasPolicyEditsMixin, BaseModel, RelatedJobsMixin):
             default='',
         )
     )
+    max_concurrent_jobs = models.IntegerField(default=0, help_text=_("Maximum number of concurrent jobs to run on this group. Zero means no limit."))
+    max_forks = models.IntegerField(default=0, help_text=_("Max forks to execute on this group. Zero means no limit."))
     policy_instance_percentage = models.IntegerField(default=0, help_text=_("Percentage of Instances to automatically assign to this group"))
     policy_instance_minimum = models.IntegerField(default=0, help_text=_("Static minimum number of Instances to automatically assign to this group"))
     policy_instance_list = JSONBlob(
@@ -391,6 +394,8 @@ class InstanceGroup(HasPolicyEditsMixin, BaseModel, RelatedJobsMixin):
 
     @property
     def capacity(self):
+        if self.is_container_group:
+            return self.max_forks
         return sum(inst.capacity for inst in self.instances.all())
 
     @property

awx/main/models/inventory.py

@@ -567,17 +567,6 @@ class Host(CommonModelNameNotUnique, RelatedJobsMixin):
     # Use .job_host_summaries.all() to get jobs affecting this host.
     # Use .job_events.all() to get events affecting this host.
 
-    '''
-    We don't use timestamp, but we may in the future.
-    '''
-
-    def update_ansible_facts(self, module, facts, timestamp=None):
-        if module == "ansible":
-            self.ansible_facts.update(facts)
-        else:
-            self.ansible_facts[module] = facts
-        self.save()
-
     def get_effective_host_name(self):
         """
         Return the name of the host that will be used in actual ansible

awx/main/models/jobs.py

@@ -44,7 +44,7 @@ from awx.main.models.notifications import (
     NotificationTemplate,
     JobNotificationMixin,
 )
-from awx.main.utils import parse_yaml_or_json, getattr_dne, NullablePromptPseudoField, polymorphic
+from awx.main.utils import parse_yaml_or_json, getattr_dne, NullablePromptPseudoField, polymorphic, log_excess_runtime
 from awx.main.fields import ImplicitRoleField, AskForField, JSONBlob, OrderedManyToManyField
 from awx.main.models.mixins import (
     ResourceMixin,
@@ -857,8 +857,11 @@ class Job(UnifiedJob, JobOptions, SurveyJobMixin, JobNotificationMixin, TaskMana
             return host_queryset.iterator()
         return host_queryset
 
-    def start_job_fact_cache(self, destination, modification_times, timeout=None):
+    @log_excess_runtime(logger, debug_cutoff=0.01, msg='Job {job_id} host facts prepared for {written_ct} hosts, took {delta:.3f} s', add_log_data=True)
+    def start_job_fact_cache(self, destination, log_data, timeout=None):
         self.log_lifecycle("start_job_fact_cache")
+        log_data['job_id'] = self.id
+        log_data['written_ct'] = 0
         os.makedirs(destination, mode=0o700)
 
         if timeout is None:
@@ -869,6 +872,8 @@ class Job(UnifiedJob, JobOptions, SurveyJobMixin, JobNotificationMixin, TaskMana
             hosts = self._get_inventory_hosts(ansible_facts_modified__gte=timeout)
         else:
             hosts = self._get_inventory_hosts()
+
+        last_filepath_written = None
         for host in hosts:
             filepath = os.sep.join(map(str, [destination, host.name]))
             if not os.path.realpath(filepath).startswith(destination):
@@ -878,23 +883,38 @@ class Job(UnifiedJob, JobOptions, SurveyJobMixin, JobNotificationMixin, TaskMana
                 with codecs.open(filepath, 'w', encoding='utf-8') as f:
                     os.chmod(f.name, 0o600)
                     json.dump(host.ansible_facts, f)
+                    log_data['written_ct'] += 1
+                    last_filepath_written = filepath
             except IOError:
                 system_tracking_logger.error('facts for host {} could not be cached'.format(smart_str(host.name)))
                 continue
-            # make note of the time we wrote the file so we can check if it changed later
-            modification_times[filepath] = os.path.getmtime(filepath)
+        # make note of the time we wrote the last file so we can check if any file changed later
+        if last_filepath_written:
+            return os.path.getmtime(last_filepath_written)
+        return None
 
-    def finish_job_fact_cache(self, destination, modification_times):
+    @log_excess_runtime(
+        logger,
+        debug_cutoff=0.01,
+        msg='Job {job_id} host facts: updated {updated_ct}, cleared {cleared_ct}, unchanged {unmodified_ct}, took {delta:.3f} s',
+        add_log_data=True,
+    )
+    def finish_job_fact_cache(self, destination, facts_write_time, log_data):
         self.log_lifecycle("finish_job_fact_cache")
+        log_data['job_id'] = self.id
+        log_data['updated_ct'] = 0
+        log_data['unmodified_ct'] = 0
+        log_data['cleared_ct'] = 0
+        hosts_to_update = []
         for host in self._get_inventory_hosts():
             filepath = os.sep.join(map(str, [destination, host.name]))
             if not os.path.realpath(filepath).startswith(destination):
                 system_tracking_logger.error('facts for host {} could not be cached'.format(smart_str(host.name)))
                 continue
             if os.path.exists(filepath):
-                # If the file changed since we wrote it pre-playbook run...
+                # If the file changed since we wrote the last facts file, pre-playbook run...
                 modified = os.path.getmtime(filepath)
-                if modified > modification_times.get(filepath, 0):
+                if (not facts_write_time) or modified > facts_write_time:
                     with codecs.open(filepath, 'r', encoding='utf-8') as f:
                         try:
                             ansible_facts = json.load(f)
@@ -902,7 +922,7 @@ class Job(UnifiedJob, JobOptions, SurveyJobMixin, JobNotificationMixin, TaskMana
                             continue
                         host.ansible_facts = ansible_facts
                         host.ansible_facts_modified = now()
-                        host.save(update_fields=['ansible_facts', 'ansible_facts_modified'])
+                        hosts_to_update.append(host)
                         system_tracking_logger.info(
                             'New fact for inventory {} host {}'.format(smart_str(host.inventory.name), smart_str(host.name)),
                             extra=dict(
@@ -913,12 +933,21 @@ class Job(UnifiedJob, JobOptions, SurveyJobMixin, JobNotificationMixin, TaskMana
                                 job_id=self.id,
                             ),
                         )
+                        log_data['updated_ct'] += 1
+                else:
+                    log_data['unmodified_ct'] += 1
             else:
                 # if the file goes missing, ansible removed it (likely via clear_facts)
                 host.ansible_facts = {}
                 host.ansible_facts_modified = now()
+                hosts_to_update.append(host)
                 system_tracking_logger.info('Facts cleared for inventory {} host {}'.format(smart_str(host.inventory.name), smart_str(host.name)))
-                host.save()
+                log_data['cleared_ct'] += 1
+            if len(hosts_to_update) > 100:
+                self.inventory.hosts.bulk_update(hosts_to_update, ['ansible_facts', 'ansible_facts_modified'])
+                hosts_to_update = []
+        if hosts_to_update:
+            self.inventory.hosts.bulk_update(hosts_to_update, ['ansible_facts', 'ansible_facts_modified'])
 
 
 class LaunchTimeConfigBase(BaseModel):

awx/main/models/unified_jobs.py

@@ -1351,12 +1351,12 @@ class UnifiedJob(
                 if required in defined_fields and not credential.has_input(required):
                     missing_credential_inputs.append(required)
 
-        if missing_credential_inputs:
-            self.job_explanation = '{} cannot start because Credential {} does not provide one or more required fields ({}).'.format(
-                self._meta.verbose_name.title(), credential.name, ', '.join(sorted(missing_credential_inputs))
-            )
-            self.save(update_fields=['job_explanation'])
-            return (False, None)
+            if missing_credential_inputs:
+                self.job_explanation = '{} cannot start because Credential {} does not provide one or more required fields ({}).'.format(
+                    self._meta.verbose_name.title(), credential.name, ', '.join(sorted(missing_credential_inputs))
+                )
+                self.save(update_fields=['job_explanation'])
+                return (False, None)
 
         needed = self.get_passwords_needed_to_start()
         try:

awx/main/routing.py

@@ -27,8 +27,8 @@ class AWXProtocolTypeRouter(ProtocolTypeRouter):
 
 
 websocket_urlpatterns = [
-    re_path(r'websocket/$', consumers.EventConsumer),
-    re_path(r'websocket/broadcast/$', consumers.BroadcastConsumer),
+    re_path(r'websocket/$', consumers.EventConsumer.as_asgi()),
+    re_path(r'websocket/broadcast/$', consumers.BroadcastConsumer.as_asgi()),
 ]
 
 application = AWXProtocolTypeRouter(

awx/main/scheduler/task_manager.py

@@ -39,12 +39,11 @@ from awx.main.utils import (
     ScheduleTaskManager,
     ScheduleWorkflowManager,
 )
-from awx.main.utils.common import task_manager_bulk_reschedule
+from awx.main.utils.common import task_manager_bulk_reschedule, is_testing
 from awx.main.signals import disable_activity_stream
 from awx.main.constants import ACTIVE_STATES
 from awx.main.scheduler.dependency_graph import DependencyGraph
-from awx.main.scheduler.task_manager_models import TaskManagerInstances
-from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups
+from awx.main.scheduler.task_manager_models import TaskManagerModels
 import awx.main.analytics.subsystem_metrics as s_metrics
 from awx.main.utils import decrypt_field
 
@@ -71,7 +70,12 @@ class TaskBase:
         # is called later.
         self.subsystem_metrics = s_metrics.Metrics(auto_pipe_execute=False)
         self.start_time = time.time()
+
+        # We want to avoid calling settings in loops, so cache these settings at init time
         self.start_task_limit = settings.START_TASK_LIMIT
+        self.task_manager_timeout = settings.TASK_MANAGER_TIMEOUT
+        self.control_task_impact = settings.AWX_CONTROL_NODE_TASK_IMPACT
+
         for m in self.subsystem_metrics.METRICS:
             if m.startswith(self.prefix):
                 self.subsystem_metrics.set(m, 0)
@@ -79,7 +83,7 @@ class TaskBase:
     def timed_out(self):
         """Return True/False if we have met or exceeded the timeout for the task manager."""
         elapsed = time.time() - self.start_time
-        if elapsed >= settings.TASK_MANAGER_TIMEOUT:
+        if elapsed >= self.task_manager_timeout:
             logger.warning(f"{self.prefix} manager has run for {elapsed} which is greater than TASK_MANAGER_TIMEOUT of {settings.TASK_MANAGER_TIMEOUT}.")
             return True
         return False
@@ -97,7 +101,7 @@ class TaskBase:
         self.all_tasks = [t for t in qs]
 
     def record_aggregate_metrics(self, *args):
-        if not settings.IS_TESTING():
+        if not is_testing():
             # increment task_manager_schedule_calls regardless if the other
             # metrics are recorded
             s_metrics.Metrics(auto_pipe_execute=True).inc(f"{self.prefix}__schedule_calls", 1)
@@ -471,9 +475,8 @@ class TaskManager(TaskBase):
         Init AFTER we know this instance of the task manager will run because the lock is acquired.
         """
         self.dependency_graph = DependencyGraph()
-        self.instances = TaskManagerInstances(self.all_tasks)
-        self.instance_groups = TaskManagerInstanceGroups(instances_by_hostname=self.instances)
-        self.controlplane_ig = self.instance_groups.controlplane_ig
+        self.tm_models = TaskManagerModels()
+        self.controlplane_ig = self.tm_models.instance_groups.controlplane_ig
 
     def job_blocked_by(self, task):
         # TODO: I'm not happy with this, I think blocking behavior should be decided outside of the dependency graph
@@ -504,8 +507,16 @@ class TaskManager(TaskBase):
         return None
 
     @timeit
-    def start_task(self, task, instance_group, dependent_tasks=None, instance=None):
+    def start_task(self, task, instance_group, instance=None):
+        # Just like for process_running_tasks, add the job to the dependency graph and
+        # ask the TaskManagerInstanceGroups object to update consumed capacity on all
+        # implicated instances and container groups.
         self.dependency_graph.add_job(task)
+        if instance_group is not None:
+            task.instance_group = instance_group
+        # We need the instance group assigned to correctly account for container group max_concurrent_jobs and max_forks
+        self.tm_models.consume_capacity(task)
+
         self.subsystem_metrics.inc(f"{self.prefix}_tasks_started", 1)
         self.start_task_limit -= 1
         if self.start_task_limit == 0:
@@ -513,20 +524,6 @@ class TaskManager(TaskBase):
             ScheduleTaskManager().schedule()
         from awx.main.tasks.system import handle_work_error, handle_work_success
 
-        # update capacity for control node and execution node
-        if task.controller_node:
-            self.instances[task.controller_node].consume_capacity(settings.AWX_CONTROL_NODE_TASK_IMPACT)
-        if task.execution_node:
-            self.instances[task.execution_node].consume_capacity(task.task_impact)
-
-        dependent_tasks = dependent_tasks or []
-
-        task_actual = {
-            'type': get_type_for_model(type(task)),
-            'id': task.id,
-        }
-        dependencies = [{'type': get_type_for_model(type(t)), 'id': t.id} for t in dependent_tasks]
-
         task.status = 'waiting'
 
         (start_status, opts) = task.pre_start()
@@ -546,7 +543,6 @@ class TaskManager(TaskBase):
                 ScheduleWorkflowManager().schedule()
             # at this point we already have control/execution nodes selected for the following cases
             else:
-                task.instance_group = instance_group
                 execution_node_msg = f' and execution node {task.execution_node}' if task.execution_node else ''
                 logger.debug(
                     f'Submitting job {task.log_format} controlled by {task.controller_node} to instance group {instance_group.name}{execution_node_msg}.'
@@ -559,6 +555,7 @@ class TaskManager(TaskBase):
         # apply_async does a NOTIFY to the channel dispatcher is listening to
         # postgres will treat this as part of the transaction, which is what we want
         if task.status != 'failed' and type(task) is not WorkflowJob:
+            task_actual = {'type': get_type_for_model(type(task)), 'id': task.id}
             task_cls = task._get_task_class()
             task_cls.apply_async(
                 [task.pk],
@@ -566,7 +563,7 @@ class TaskManager(TaskBase):
                 queue=task.get_queue_name(),
                 uuid=task.celery_task_id,
                 callbacks=[{'task': handle_work_success.name, 'kwargs': {'task_actual': task_actual}}],
-                errbacks=[{'task': handle_work_error.name, 'args': [task.celery_task_id], 'kwargs': {'subtasks': [task_actual] + dependencies}}],
+                errbacks=[{'task': handle_work_error.name, 'kwargs': {'task_actual': task_actual}}],
             )
 
         # In exception cases, like a job failing pre-start checks, we send the websocket status message
@@ -580,6 +577,7 @@ class TaskManager(TaskBase):
             if type(task) is WorkflowJob:
                 ScheduleWorkflowManager().schedule()
             self.dependency_graph.add_job(task)
+            self.tm_models.consume_capacity(task)
 
     @timeit
     def process_pending_tasks(self, pending_tasks):
@@ -604,18 +602,18 @@ class TaskManager(TaskBase):
             if isinstance(task, WorkflowJob):
                 # Previously we were tracking allow_simultaneous blocking both here and in DependencyGraph.
                 # Double check that using just the DependencyGraph works for Workflows and Sliced Jobs.
-                self.start_task(task, None, task.get_jobs_fail_chain(), None)
+                self.start_task(task, None, None)
                 continue
 
             found_acceptable_queue = False
 
             # Determine if there is control capacity for the task
             if task.capacity_type == 'control':
-                control_impact = task.task_impact + settings.AWX_CONTROL_NODE_TASK_IMPACT
+                control_impact = task.task_impact + self.control_task_impact
             else:
-                control_impact = settings.AWX_CONTROL_NODE_TASK_IMPACT
-            control_instance = self.instance_groups.fit_task_to_most_remaining_capacity_instance(
-                task, instance_group_name=settings.DEFAULT_CONTROL_PLANE_QUEUE_NAME, impact=control_impact, capacity_type='control'
+                control_impact = self.control_task_impact
+            control_instance = self.tm_models.instance_groups.fit_task_to_most_remaining_capacity_instance(
+                task, instance_group_name=self.controlplane_ig.name, impact=control_impact, capacity_type='control'
             )
             if not control_instance:
                 self.task_needs_capacity(task, tasks_to_update_job_explanation)
@@ -626,25 +624,29 @@ class TaskManager(TaskBase):
 
             # All task.capacity_type == 'control' jobs should run on control plane, no need to loop over instance groups
             if task.capacity_type == 'control':
+                if not self.tm_models.instance_groups[self.controlplane_ig.name].has_remaining_capacity(control_impact=True):
+                    continue
                 task.execution_node = control_instance.hostname
-                execution_instance = self.instances[control_instance.hostname].obj
+                execution_instance = self.tm_models.instances[control_instance.hostname].obj
                 task.log_lifecycle("controller_node_chosen")
                 task.log_lifecycle("execution_node_chosen")
-                self.start_task(task, self.controlplane_ig, task.get_jobs_fail_chain(), execution_instance)
+                self.start_task(task, self.controlplane_ig, execution_instance)
                 found_acceptable_queue = True
                 continue
 
-            for instance_group in self.instance_groups.get_instance_groups_from_task_cache(task):
+            for instance_group in self.tm_models.instance_groups.get_instance_groups_from_task_cache(task):
+                if not self.tm_models.instance_groups[instance_group.name].has_remaining_capacity(task):
+                    continue
                 if instance_group.is_container_group:
-                    self.start_task(task, instance_group, task.get_jobs_fail_chain(), None)
+                    self.start_task(task, instance_group, None)
                     found_acceptable_queue = True
                     break
 
                 # at this point we know the instance group is NOT a container group
                 # because if it was, it would have started the task and broke out of the loop.
-                execution_instance = self.instance_groups.fit_task_to_most_remaining_capacity_instance(
+                execution_instance = self.tm_models.instance_groups.fit_task_to_most_remaining_capacity_instance(
                     task, instance_group_name=instance_group.name, add_hybrid_control_cost=True
-                ) or self.instance_groups.find_largest_idle_instance(instance_group_name=instance_group.name, capacity_type=task.capacity_type)
+                ) or self.tm_models.instance_groups.find_largest_idle_instance(instance_group_name=instance_group.name, capacity_type=task.capacity_type)
 
                 if execution_instance:
                     task.execution_node = execution_instance.hostname
@@ -660,8 +662,8 @@ class TaskManager(TaskBase):
                             task.log_format, instance_group.name, execution_instance.hostname, execution_instance.remaining_capacity
                         )
                     )
-                    execution_instance = self.instances[execution_instance.hostname].obj
-                    self.start_task(task, instance_group, task.get_jobs_fail_chain(), execution_instance)
+                    execution_instance = self.tm_models.instances[execution_instance.hostname].obj
+                    self.start_task(task, instance_group, execution_instance)
                     found_acceptable_queue = True
                     break
                 else:

awx/main/scheduler/task_manager_models.py

@@ -15,15 +15,18 @@ logger = logging.getLogger('awx.main.scheduler')
 class TaskManagerInstance:
     """A class representing minimal data the task manager needs to represent an Instance."""
 
-    def __init__(self, obj):
+    def __init__(self, obj, **kwargs):
         self.obj = obj
         self.node_type = obj.node_type
         self.consumed_capacity = 0
         self.capacity = obj.capacity
         self.hostname = obj.hostname
+        self.jobs_running = 0
 
-    def consume_capacity(self, impact):
+    def consume_capacity(self, impact, job_impact=False):
         self.consumed_capacity += impact
+        if job_impact:
+            self.jobs_running += 1
 
     @property
     def remaining_capacity(self):
@@ -33,9 +36,106 @@ class TaskManagerInstance:
         return remaining
 
 
+class TaskManagerInstanceGroup:
+    """A class representing minimal data the task manager needs to represent an InstanceGroup."""
+
+    def __init__(self, obj, task_manager_instances=None, **kwargs):
+        self.name = obj.name
+        self.is_container_group = obj.is_container_group
+        self.container_group_jobs = 0
+        self.container_group_consumed_forks = 0
+        _instances = obj.instances.all()
+        # We want the list of TaskManagerInstance objects because these are shared across the TaskManagerInstanceGroup objects.
+        # This way when we consume capacity on an instance that is in multiple groups, we tabulate across all the groups correctly.
+        self.instances = [task_manager_instances[instance.hostname] for instance in _instances if instance.hostname in task_manager_instances]
+        self.instance_hostnames = tuple([instance.hostname for instance in _instances if instance.hostname in task_manager_instances])
+        self.max_concurrent_jobs = obj.max_concurrent_jobs
+        self.max_forks = obj.max_forks
+        self.control_task_impact = kwargs.get('control_task_impact', settings.AWX_CONTROL_NODE_TASK_IMPACT)
+
+    def consume_capacity(self, task):
+        """We only consume capacity on an instance group level if it is a container group. Otherwise we consume capacity on an instance level."""
+        if self.is_container_group:
+            self.container_group_jobs += 1
+            self.container_group_consumed_forks += task.task_impact
+        else:
+            raise RuntimeError("We only track capacity for container groups at the instance group level. Otherwise, consume capacity on instances.")
+
+    def get_remaining_instance_capacity(self):
+        return sum(inst.remaining_capacity for inst in self.instances)
+
+    def get_instance_capacity(self):
+        return sum(inst.capacity for inst in self.instances)
+
+    def get_consumed_instance_capacity(self):
+        return sum(inst.consumed_capacity for inst in self.instances)
+
+    def get_instance_jobs_running(self):
+        return sum(inst.jobs_running for inst in self.instances)
+
+    def get_jobs_running(self):
+        if self.is_container_group:
+            return self.container_group_jobs
+        return sum(inst.jobs_running for inst in self.instances)
+
+    def get_capacity(self):
+        """This reports any type of capacity, including that of container group jobs.
+
+        Container groups don't really have capacity, but if they have max_forks set,
+        we can interperet that as how much capacity the user has defined them to have.
+        """
+        if self.is_container_group:
+            return self.max_forks
+        return self.get_instance_capacity()
+
+    def get_consumed_capacity(self):
+        if self.is_container_group:
+            return self.container_group_consumed_forks
+        return self.get_consumed_instance_capacity()
+
+    def get_remaining_capacity(self):
+        return self.get_capacity() - self.get_consumed_capacity()
+
+    def has_remaining_capacity(self, task=None, control_impact=False):
+        """Pass either a task or control_impact=True to determine if the IG has capacity to run the control task or job task."""
+        task_impact = self.control_task_impact if control_impact else task.task_impact
+        job_impact = 0 if control_impact else 1
+        task_string = f"task {task.log_format} with impact of {task_impact}" if task else f"control task with impact of {task_impact}"
+
+        # We only want to loop over instances if self.max_concurrent_jobs is set
+        if self.max_concurrent_jobs == 0:
+            # Override the calculated remaining capacity, because when max_concurrent_jobs == 0 we don't enforce any max
+            remaining_jobs = 0
+        else:
+            remaining_jobs = self.max_concurrent_jobs - self.get_jobs_running() - job_impact
+
+        # We only want to loop over instances if self.max_forks is set
+        if self.max_forks == 0:
+            # Override the calculated remaining capacity, because when max_forks == 0 we don't enforce any max
+            remaining_forks = 0
+        else:
+            remaining_forks = self.max_forks - self.get_consumed_capacity() - task_impact
+
+        if remaining_jobs < 0 or remaining_forks < 0:
+            # A value less than zero means the task will not fit on the group
+            if remaining_jobs < 0:
+                logger.debug(f"{task_string} cannot fit on instance group {self.name} with {remaining_jobs} remaining jobs")
+            if remaining_forks < 0:
+                logger.debug(f"{task_string} cannot fit on instance group {self.name} with {remaining_forks} remaining forks")
+            return False
+
+        # Returning true means there is enough remaining capacity on the group to run the task (or no instance group level limits are being set)
+        logger.debug(f"{task_string} can fit on instance group {self.name} with {remaining_forks} remaining forks and {remaining_jobs}")
+        return True
+
+
 class TaskManagerInstances:
-    def __init__(self, active_tasks, instances=None, instance_fields=('node_type', 'capacity', 'hostname', 'enabled')):
+    def __init__(self, instances=None, instance_fields=('node_type', 'capacity', 'hostname', 'enabled'), **kwargs):
         self.instances_by_hostname = dict()
+        self.instance_groups_container_group_jobs = dict()
+        self.instance_groups_container_group_consumed_forks = dict()
+        self.control_task_impact = kwargs.get('control_task_impact', settings.AWX_CONTROL_NODE_TASK_IMPACT)
+
         if instances is None:
             instances = (
                 Instance.objects.filter(hostname__isnull=False, node_state=Instance.States.READY, enabled=True)
@@ -43,18 +143,15 @@ class TaskManagerInstances:
                 .only('node_type', 'node_state', 'capacity', 'hostname', 'enabled')
             )
         for instance in instances:
-            self.instances_by_hostname[instance.hostname] = TaskManagerInstance(instance)
+            self.instances_by_hostname[instance.hostname] = TaskManagerInstance(instance, **kwargs)
 
-        # initialize remaining capacity based on currently waiting and running tasks
-        for task in active_tasks:
-            if task.status not in ['waiting', 'running']:
-                continue
-            control_instance = self.instances_by_hostname.get(task.controller_node, '')
-            execution_instance = self.instances_by_hostname.get(task.execution_node, '')
-            if execution_instance and execution_instance.node_type in ('hybrid', 'execution'):
-                self.instances_by_hostname[task.execution_node].consume_capacity(task.task_impact)
-            if control_instance and control_instance.node_type in ('hybrid', 'control'):
-                self.instances_by_hostname[task.controller_node].consume_capacity(settings.AWX_CONTROL_NODE_TASK_IMPACT)
+    def consume_capacity(self, task):
+        control_instance = self.instances_by_hostname.get(task.controller_node, '')
+        execution_instance = self.instances_by_hostname.get(task.execution_node, '')
+        if execution_instance and execution_instance.node_type in ('hybrid', 'execution'):
+            self.instances_by_hostname[task.execution_node].consume_capacity(task.task_impact, job_impact=True)
+        if control_instance and control_instance.node_type in ('hybrid', 'control'):
+            self.instances_by_hostname[task.controller_node].consume_capacity(self.control_task_impact)
 
     def __getitem__(self, hostname):
         return self.instances_by_hostname.get(hostname)
@@ -64,42 +161,57 @@ class TaskManagerInstances:
 
 
 class TaskManagerInstanceGroups:
-    """A class representing minimal data the task manager needs to represent an InstanceGroup."""
+    """A class representing minimal data the task manager needs to represent all the InstanceGroups."""
 
-    def __init__(self, instances_by_hostname=None, instance_groups=None, instance_groups_queryset=None):
+    def __init__(self, task_manager_instances=None, instance_groups=None, instance_groups_queryset=None, **kwargs):
         self.instance_groups = dict()
+        self.task_manager_instances = task_manager_instances if task_manager_instances is not None else TaskManagerInstances()
         self.controlplane_ig = None
         self.pk_ig_map = dict()
+        self.control_task_impact = kwargs.get('control_task_impact', settings.AWX_CONTROL_NODE_TASK_IMPACT)
+        self.controlplane_ig_name = kwargs.get('controlplane_ig_name', settings.DEFAULT_CONTROL_PLANE_QUEUE_NAME)
 
         if instance_groups is not None:  # for testing
-            self.instance_groups = instance_groups
+            self.instance_groups = {ig.name: TaskManagerInstanceGroup(ig, self.task_manager_instances, **kwargs) for ig in instance_groups}
+            self.pk_ig_map = {ig.pk: ig for ig in instance_groups}
         else:
             if instance_groups_queryset is None:
-                instance_groups_queryset = InstanceGroup.objects.prefetch_related('instances').only('name', 'instances')
-            for instance_group in instance_groups_queryset:
-                if instance_group.name == settings.DEFAULT_CONTROL_PLANE_QUEUE_NAME:
-                    self.controlplane_ig = instance_group
-                self.instance_groups[instance_group.name] = dict(
-                    instances=[
-                        instances_by_hostname[instance.hostname] for instance in instance_group.instances.all() if instance.hostname in instances_by_hostname
-                    ],
+                instance_groups_queryset = InstanceGroup.objects.prefetch_related('instances').only(
+                    'name', 'instances', 'max_concurrent_jobs', 'max_forks', 'is_container_group'
                 )
+            for instance_group in instance_groups_queryset:
+                if instance_group.name == self.controlplane_ig_name:
+                    self.controlplane_ig = instance_group
+                self.instance_groups[instance_group.name] = TaskManagerInstanceGroup(instance_group, self.task_manager_instances, **kwargs)
                 self.pk_ig_map[instance_group.pk] = instance_group
 
+    def __getitem__(self, ig_name):
+        return self.instance_groups.get(ig_name)
+
+    def __contains__(self, ig_name):
+        return ig_name in self.instance_groups
+
     def get_remaining_capacity(self, group_name):
-        instances = self.instance_groups[group_name]['instances']
-        return sum(inst.remaining_capacity for inst in instances)
+        return self.instance_groups[group_name].get_remaining_instance_capacity()
 
     def get_consumed_capacity(self, group_name):
-        instances = self.instance_groups[group_name]['instances']
-        return sum(inst.consumed_capacity for inst in instances)
+        return self.instance_groups[group_name].get_consumed_capacity()
+
+    def get_jobs_running(self, group_name):
+        return self.instance_groups[group_name].get_jobs_running()
+
+    def get_capacity(self, group_name):
+        return self.instance_groups[group_name].get_capacity()
+
+    def get_instances(self, group_name):
+        return self.instance_groups[group_name].instances
 
     def fit_task_to_most_remaining_capacity_instance(self, task, instance_group_name, impact=None, capacity_type=None, add_hybrid_control_cost=False):
         impact = impact if impact else task.task_impact
         capacity_type = capacity_type if capacity_type else task.capacity_type
         instance_most_capacity = None
         most_remaining_capacity = -1
-        instances = self.instance_groups[instance_group_name]['instances']
+        instances = self.instance_groups[instance_group_name].instances
 
         for i in instances:
             if i.node_type not in (capacity_type, 'hybrid'):
@@ -107,7 +219,7 @@ class TaskManagerInstanceGroups:
             would_be_remaining = i.remaining_capacity - impact
             # hybrid nodes _always_ control their own tasks
             if add_hybrid_control_cost and i.node_type == 'hybrid':
-                would_be_remaining -= settings.AWX_CONTROL_NODE_TASK_IMPACT
+                would_be_remaining -= self.control_task_impact
             if would_be_remaining >= 0 and (instance_most_capacity is None or would_be_remaining > most_remaining_capacity):
                 instance_most_capacity = i
                 most_remaining_capacity = would_be_remaining
@@ -115,10 +227,13 @@ class TaskManagerInstanceGroups:
 
     def find_largest_idle_instance(self, instance_group_name, capacity_type='execution'):
         largest_instance = None
-        instances = self.instance_groups[instance_group_name]['instances']
+        instances = self.instance_groups[instance_group_name].instances
         for i in instances:
             if i.node_type not in (capacity_type, 'hybrid'):
                 continue
+            if i.capacity <= 0:
+                # We don't want to select an idle instance with 0 capacity
+                continue
             if (hasattr(i, 'jobs_running') and i.jobs_running == 0) or i.remaining_capacity == i.capacity:
                 if largest_instance is None:
                     largest_instance = i
@@ -139,3 +254,56 @@ class TaskManagerInstanceGroups:
             logger.warn(f"No instance groups in cache exist, defaulting to global instance groups for task {task}")
             return task.global_instance_groups
         return igs
+
+
+class TaskManagerModels:
+    def __init__(self, **kwargs):
+        # We want to avoid calls to settings over and over in loops, so cache this information here
+        kwargs['control_task_impact'] = kwargs.get('control_task_impact', settings.AWX_CONTROL_NODE_TASK_IMPACT)
+        kwargs['controlplane_ig_name'] = kwargs.get('controlplane_ig_name', settings.DEFAULT_CONTROL_PLANE_QUEUE_NAME)
+        self.instances = TaskManagerInstances(**kwargs)
+        self.instance_groups = TaskManagerInstanceGroups(task_manager_instances=self.instances, **kwargs)
+
+    @classmethod
+    def init_with_consumed_capacity(cls, **kwargs):
+        tmm = cls(**kwargs)
+        tasks = kwargs.get('tasks', None)
+
+        if tasks is None:
+            instance_group_queryset = kwargs.get('instance_groups_queryset', None)
+            # No tasks were provided, so we will fetch them from the database
+            task_status_filter_list = kwargs.get('task_status_filter_list', ['running', 'waiting'])
+            task_fields = kwargs.get('task_fields', ('task_impact', 'controller_node', 'execution_node', 'instance_group'))
+            from awx.main.models import UnifiedJob
+
+            if instance_group_queryset is not None:
+                logger.debug("******************INSTANCE GROUP QUERYSET PASSED -- FILTERING TASKS ****************************")
+                # Sometimes things like the serializer pass a queryset that looks at not all instance groups. in this case,
+                # we also need to filter the tasks we look at
+                tasks = UnifiedJob.objects.filter(status__in=task_status_filter_list, instance_group__in=[ig.id for ig in instance_group_queryset]).only(
+                    *task_fields
+                )
+            else:
+                # No instance group query set, look at all tasks in whole system
+                tasks = UnifiedJob.objects.filter(status__in=task_status_filter_list).only(*task_fields)
+
+        for task in tasks:
+            tmm.consume_capacity(task)
+
+        return tmm
+
+    def consume_capacity(self, task):
+        # Consume capacity on instances, which bubbles up to instance groups they are a member of
+        self.instances.consume_capacity(task)
+
+        # For container group jobs, additionally we must account for capacity consumed since
+        # The container groups have no instances to look at to track how many jobs/forks are consumed
+        if task.instance_group_id:
+            if not task.instance_group_id in self.instance_groups.pk_ig_map.keys():
+                logger.warn(
+                    f"Task {task.log_format} assigned {task.instance_group_id} but this instance group not present in map of instance groups{self.instance_groups.pk_ig_map.keys()}"
+                )
+            else:
+                ig = self.instance_groups.pk_ig_map[task.instance_group_id]
+                if ig.is_container_group:
+                    self.instance_groups[ig.name].consume_capacity(task)

awx/main/tasks/jobs.py

@@ -426,7 +426,7 @@ class BaseTask(object):
         """
         instance.log_lifecycle("post_run")
 
-    def final_run_hook(self, instance, status, private_data_dir, fact_modification_times):
+    def final_run_hook(self, instance, status, private_data_dir):
         """
         Hook for any steps to run after job/task is marked as complete.
         """
@@ -469,7 +469,6 @@ class BaseTask(object):
         self.instance = self.update_model(pk, status='running', start_args='')  # blank field to remove encrypted passwords
         self.instance.websocket_emit_status("running")
         status, rc = 'error', None
-        fact_modification_times = {}
         self.runner_callback.event_ct = 0
 
         '''
@@ -498,14 +497,6 @@ class BaseTask(object):
             if not os.path.exists(settings.AWX_ISOLATION_BASE_PATH):
                 raise RuntimeError('AWX_ISOLATION_BASE_PATH=%s does not exist' % settings.AWX_ISOLATION_BASE_PATH)
 
-            # Fetch "cached" fact data from prior runs and put on the disk
-            # where ansible expects to find it
-            if getattr(self.instance, 'use_fact_cache', False):
-                self.instance.start_job_fact_cache(
-                    os.path.join(private_data_dir, 'artifacts', str(self.instance.id), 'fact_cache'),
-                    fact_modification_times,
-                )
-
             # May have to serialize the value
             private_data_files, ssh_key_data = self.build_private_data_files(self.instance, private_data_dir)
             passwords = self.build_passwords(self.instance, kwargs)
@@ -646,7 +637,7 @@ class BaseTask(object):
             self.instance.send_notification_templates('succeeded' if status == 'successful' else 'failed')
 
         try:
-            self.final_run_hook(self.instance, status, private_data_dir, fact_modification_times)
+            self.final_run_hook(self.instance, status, private_data_dir)
         except Exception:
             logger.exception('{} Final run hook errored.'.format(self.instance.log_format))
 
@@ -1066,12 +1057,19 @@ class RunJob(SourceControlMixin, BaseTask):
             # ran inside of the event saving code
             update_smart_memberships_for_inventory(job.inventory)
 
+        # Fetch "cached" fact data from prior runs and put on the disk
+        # where ansible expects to find it
+        if job.use_fact_cache:
+            self.facts_write_time = self.instance.start_job_fact_cache(os.path.join(private_data_dir, 'artifacts', str(job.id), 'fact_cache'))
+
     def build_project_dir(self, job, private_data_dir):
         self.sync_and_copy(job.project, private_data_dir, scm_branch=job.scm_branch)
 
-    def final_run_hook(self, job, status, private_data_dir, fact_modification_times):
-        super(RunJob, self).final_run_hook(job, status, private_data_dir, fact_modification_times)
-        if not private_data_dir:
+    def post_run_hook(self, job, status):
+        super(RunJob, self).post_run_hook(job, status)
+        job.refresh_from_db(fields=['job_env'])
+        private_data_dir = job.job_env.get('AWX_PRIVATE_DATA_DIR')
+        if (not private_data_dir) or (not hasattr(self, 'facts_write_time')):
             # If there's no private data dir, that means we didn't get into the
             # actual `run()` call; this _usually_ means something failed in
             # the pre_run_hook method
@@ -1079,9 +1077,11 @@ class RunJob(SourceControlMixin, BaseTask):
         if job.use_fact_cache:
             job.finish_job_fact_cache(
                 os.path.join(private_data_dir, 'artifacts', str(job.id), 'fact_cache'),
-                fact_modification_times,
+                self.facts_write_time,
             )
 
+    def final_run_hook(self, job, status, private_data_dir):
+        super(RunJob, self).final_run_hook(job, status, private_data_dir)
         try:
             inventory = job.inventory
         except Inventory.DoesNotExist:

awx/main/tasks/receptor.py

@@ -61,10 +61,15 @@ def read_receptor_config():
             return yaml.safe_load(f)
 
 
-def get_receptor_sockfile():
-    data = read_receptor_config()
+def work_signing_enabled(config_data):
+    for section in config_data:
+        if 'work-signing' in section:
+            return True
+    return False
 
-    for section in data:
+
+def get_receptor_sockfile(config_data):
+    for section in config_data:
         for entry_name, entry_data in section.items():
             if entry_name == 'control-service':
                 if 'filename' in entry_data:
@@ -75,12 +80,11 @@ def get_receptor_sockfile():
         raise RuntimeError(f'Receptor conf {__RECEPTOR_CONF} does not have control-service entry needed to get sockfile')
 
 
-def get_tls_client(use_stream_tls=None):
+def get_tls_client(config_data, use_stream_tls=None):
     if not use_stream_tls:
         return None
 
-    data = read_receptor_config()
-    for section in data:
+    for section in config_data:
         for entry_name, entry_data in section.items():
             if entry_name == 'tls-client':
                 if 'name' in entry_data:
@@ -88,10 +92,12 @@ def get_tls_client(use_stream_tls=None):
     return None
 
 
-def get_receptor_ctl():
-    receptor_sockfile = get_receptor_sockfile()
+def get_receptor_ctl(config_data=None):
+    if config_data is None:
+        config_data = read_receptor_config()
+    receptor_sockfile = get_receptor_sockfile(config_data)
     try:
-        return ReceptorControl(receptor_sockfile, config=__RECEPTOR_CONF, tlsclient=get_tls_client(True))
+        return ReceptorControl(receptor_sockfile, config=__RECEPTOR_CONF, tlsclient=get_tls_client(config_data, True))
     except RuntimeError:
         return ReceptorControl(receptor_sockfile)
 
@@ -159,15 +165,18 @@ def run_until_complete(node, timing_data=None, **kwargs):
     """
     Runs an ansible-runner work_type on remote node, waits until it completes, then returns stdout.
     """
-    receptor_ctl = get_receptor_ctl()
+    config_data = read_receptor_config()
+    receptor_ctl = get_receptor_ctl(config_data)
 
     use_stream_tls = getattr(get_conn_type(node, receptor_ctl), 'name', None) == "STREAMTLS"
-    kwargs.setdefault('tlsclient', get_tls_client(use_stream_tls))
+    kwargs.setdefault('tlsclient', get_tls_client(config_data, use_stream_tls))
     kwargs.setdefault('ttl', '20s')
     kwargs.setdefault('payload', '')
+    if work_signing_enabled(config_data):
+        kwargs['signwork'] = True
 
     transmit_start = time.time()
-    result = receptor_ctl.submit_work(worktype='ansible-runner', node=node, signwork=True, **kwargs)
+    result = receptor_ctl.submit_work(worktype='ansible-runner', node=node, **kwargs)
 
     unit_id = result['unitid']
     run_start = time.time()
@@ -302,7 +311,8 @@ class AWXReceptorJob:
 
     def run(self):
         # We establish a connection to the Receptor socket
-        receptor_ctl = get_receptor_ctl()
+        self.config_data = read_receptor_config()
+        receptor_ctl = get_receptor_ctl(self.config_data)
 
         res = None
         try:
@@ -327,7 +337,7 @@ class AWXReceptorJob:
         if self.work_type == 'ansible-runner':
             work_submit_kw['node'] = self.task.instance.execution_node
             use_stream_tls = get_conn_type(work_submit_kw['node'], receptor_ctl).name == "STREAMTLS"
-            work_submit_kw['tlsclient'] = get_tls_client(use_stream_tls)
+            work_submit_kw['tlsclient'] = get_tls_client(self.config_data, use_stream_tls)
 
         with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor:
             transmitter_future = executor.submit(self.transmit, sockin)
@@ -477,7 +487,9 @@ class AWXReceptorJob:
 
     @property
     def sign_work(self):
-        return True if self.work_type in ('ansible-runner', 'local') else False
+        if self.work_type in ('ansible-runner', 'local'):
+            return work_signing_enabled(self.config_data)
+        return False
 
     @property
     def work_type(self):

awx/main/tasks/system.py

@@ -52,6 +52,7 @@ from awx.main.constants import ACTIVE_STATES
 from awx.main.dispatch.publish import task
 from awx.main.dispatch import get_local_queuename, reaper
 from awx.main.utils.common import (
+    get_type_for_model,
     ignore_inventory_computed_fields,
     ignore_inventory_group_removal,
     ScheduleWorkflowManager,
@@ -720,45 +721,43 @@ def handle_work_success(task_actual):
 
 
 @task(queue=get_local_queuename)
-def handle_work_error(task_id, *args, **kwargs):
-    subtasks = kwargs.get('subtasks', None)
-    logger.debug('Executing error task id %s, subtasks: %s' % (task_id, str(subtasks)))
-    first_instance = None
-    first_instance_type = ''
-    if subtasks is not None:
-        for each_task in subtasks:
-            try:
-                instance = UnifiedJob.get_instance_by_type(each_task['type'], each_task['id'])
-                if not instance:
-                    # Unknown task type
-                    logger.warning("Unknown task type: {}".format(each_task['type']))
-                    continue
-            except ObjectDoesNotExist:
-                logger.warning('Missing {} `{}` in error callback.'.format(each_task['type'], each_task['id']))
-                continue
+def handle_work_error(task_actual):
+    try:
+        instance = UnifiedJob.get_instance_by_type(task_actual['type'], task_actual['id'])
+    except ObjectDoesNotExist:
+        logger.warning('Missing {} `{}` in error callback.'.format(task_actual['type'], task_actual['id']))
+        return
+    if not instance:
+        return
 
-            if first_instance is None:
-                first_instance = instance
-                first_instance_type = each_task['type']
+    subtasks = instance.get_jobs_fail_chain()  # reverse of dependent_jobs mostly
+    logger.debug(f'Executing error task id {task_actual["id"]}, subtasks: {[subtask.id for subtask in subtasks]}')
 
-            if instance.celery_task_id != task_id and not instance.cancel_flag and not instance.status in ('successful', 'failed'):
-                instance.status = 'failed'
-                instance.failed = True
-                if not instance.job_explanation:
-                    instance.job_explanation = 'Previous Task Failed: {"job_type": "%s", "job_name": "%s", "job_id": "%s"}' % (
-                        first_instance_type,
-                        first_instance.name,
-                        first_instance.id,
-                    )
-                instance.save()
-                instance.websocket_emit_status("failed")
+    deps_of_deps = {}
+
+    for subtask in subtasks:
+        if subtask.celery_task_id != instance.celery_task_id and not subtask.cancel_flag and not subtask.status in ('successful', 'failed'):
+            # If there are multiple in the dependency chain, A->B->C, and this was called for A, blame B for clarity
+            blame_job = deps_of_deps.get(subtask.id, instance)
+            subtask.status = 'failed'
+            subtask.failed = True
+            if not subtask.job_explanation:
+                subtask.job_explanation = 'Previous Task Failed: {"job_type": "%s", "job_name": "%s", "job_id": "%s"}' % (
+                    get_type_for_model(type(blame_job)),
+                    blame_job.name,
+                    blame_job.id,
+                )
+            subtask.save()
+            subtask.websocket_emit_status("failed")
+
+            for sub_subtask in subtask.get_jobs_fail_chain():
+                deps_of_deps[sub_subtask.id] = subtask
 
     # We only send 1 job complete message since all the job completion message
     # handling does is trigger the scheduler. If we extend the functionality of
     # what the job complete message handler does then we may want to send a
     # completion event for each job here.
-    if first_instance:
-        schedule_manager_success_or_error(first_instance)
+    schedule_manager_success_or_error(instance)
 
 
 @task(queue=get_local_queuename)

awx/main/tests/data/inventory/plugins/ec2/env.json

@@ -3,5 +3,6 @@
     "ANSIBLE_TRANSFORM_INVALID_GROUP_CHARS": "never",
     "AWS_ACCESS_KEY_ID": "fooo",
     "AWS_SECRET_ACCESS_KEY": "fooo",
-    "AWS_SECURITY_TOKEN": "fooo"
+    "AWS_SECURITY_TOKEN": "fooo",
+    "AWS_SESSION_TOKEN": "fooo"
 }

awx/main/tests/functional/commands/test_secret_key_regeneration.py

@@ -171,13 +171,17 @@ class TestKeyRegeneration:
 
     def test_use_custom_key_with_empty_tower_secret_key_env_var(self):
         os.environ['TOWER_SECRET_KEY'] = ''
-        new_key = call_command('regenerate_secret_key', '--use-custom-key')
-        assert settings.SECRET_KEY != new_key
+        with pytest.raises(SystemExit) as e:
+            call_command('regenerate_secret_key', '--use-custom-key')
+        assert e.type == SystemExit
+        assert e.value.code == 1
 
     def test_use_custom_key_with_no_tower_secret_key_env_var(self):
         os.environ.pop('TOWER_SECRET_KEY', None)
-        new_key = call_command('regenerate_secret_key', '--use-custom-key')
-        assert settings.SECRET_KEY != new_key
+        with pytest.raises(SystemExit) as e:
+            call_command('regenerate_secret_key', '--use-custom-key')
+        assert e.type == SystemExit
+        assert e.value.code == 1
 
     def test_with_tower_secret_key_env_var(self):
         custom_key = 'MXSq9uqcwezBOChl/UfmbW1k4op+bC+FQtwPqgJ1u9XV'

awx/main/tests/functional/task_management/test_capacity.py

@@ -4,7 +4,7 @@ from awx.main.models import (
     Instance,
     InstanceGroup,
 )
-from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups, TaskManagerInstances
+from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups
 
 
 class TestInstanceGroupInstanceMapping(TransactionTestCase):
@@ -23,11 +23,10 @@ class TestInstanceGroupInstanceMapping(TransactionTestCase):
     def test_mapping(self):
         self.sample_cluster()
         with self.assertNumQueries(3):
-            instances = TaskManagerInstances([])  # empty task list
-            instance_groups = TaskManagerInstanceGroups(instances_by_hostname=instances)
+            instance_groups = TaskManagerInstanceGroups()
 
         ig_instance_map = instance_groups.instance_groups
 
-        assert set(i.hostname for i in ig_instance_map['ig_small']['instances']) == set(['i1'])
-        assert set(i.hostname for i in ig_instance_map['ig_large']['instances']) == set(['i2', 'i3'])
-        assert set(i.hostname for i in ig_instance_map['default']['instances']) == set(['i2'])
+        assert set(i.hostname for i in ig_instance_map['ig_small'].instances) == set(['i1'])
+        assert set(i.hostname for i in ig_instance_map['ig_large'].instances) == set(['i2', 'i3'])
+        assert set(i.hostname for i in ig_instance_map['default'].instances) == set(['i2'])

awx/main/tests/functional/task_management/test_container_groups.py

@@ -10,6 +10,10 @@ from awx.main.utils import (
     create_temporary_fifo,
 )
 
+from awx.main.scheduler import TaskManager
+
+from . import create_job
+
 
 @pytest.fixture
 def containerized_job(default_instance_group, kube_credential, job_template_factory):
@@ -34,6 +38,50 @@ def test_containerized_job(containerized_job):
     assert containerized_job.instance_group.credential.kubernetes
 
 
+@pytest.mark.django_db
+def test_max_concurrent_jobs_blocks_start_of_new_jobs(controlplane_instance_group, containerized_job, mocker):
+    """Construct a scenario where only 1 job will fit within the max_concurrent_jobs of the container group.
+
+    Since max_concurrent_jobs is set to 1, even though 2 jobs are in pending
+    and would be launched into the container group, only one will be started.
+    """
+    containerized_job.unified_job_template.allow_simultaneous = True
+    containerized_job.unified_job_template.save()
+    default_instance_group = containerized_job.instance_group
+    default_instance_group.max_concurrent_jobs = 1
+    default_instance_group.save()
+    task_impact = 1
+    # Create a second job that should not be scheduled at first, blocked by the other
+    create_job(containerized_job.unified_job_template)
+    tm = TaskManager()
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = task_impact
+        with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
+            tm.schedule()
+            mock_job.assert_called_once()
+
+
+@pytest.mark.django_db
+def test_max_forks_blocks_start_of_new_jobs(controlplane_instance_group, containerized_job, mocker):
+    """Construct a scenario where only 1 job will fit within the max_forks of the container group.
+
+    In this case, we set the container_group max_forks to 10, and make the task_impact of a job 6.
+    Therefore, only 1 job will fit within the max of 10.
+    """
+    containerized_job.unified_job_template.allow_simultaneous = True
+    containerized_job.unified_job_template.save()
+    default_instance_group = containerized_job.instance_group
+    default_instance_group.max_forks = 10
+    # Create a second job that should not be scheduled
+    create_job(containerized_job.unified_job_template)
+    tm = TaskManager()
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = 6
+        with mock.patch("awx.main.scheduler.TaskManager.start_task"):
+            tm.schedule()
+            tm.start_task.assert_called_once()
+
+
 @pytest.mark.django_db
 def test_kubectl_ssl_verification(containerized_job, default_job_execution_environment):
     containerized_job.execution_environment = default_job_execution_environment

awx/main/tests/functional/task_management/test_rampart_groups.py

@@ -23,7 +23,7 @@ def test_multi_group_basic_job_launch(instance_factory, controlplane_instance_gr
         mock_task_impact.return_value = 500
         with mocker.patch("awx.main.scheduler.TaskManager.start_task"):
             TaskManager().schedule()
-            TaskManager.start_task.assert_has_calls([mock.call(j1, ig1, [], i1), mock.call(j2, ig2, [], i2)])
+            TaskManager.start_task.assert_has_calls([mock.call(j1, ig1, i1), mock.call(j2, ig2, i2)])
 
 
 @pytest.mark.django_db
@@ -54,7 +54,7 @@ def test_multi_group_with_shared_dependency(instance_factory, controlplane_insta
         DependencyManager().schedule()
         TaskManager().schedule()
         pu = p.project_updates.first()
-        TaskManager.start_task.assert_called_once_with(pu, controlplane_instance_group, [j1, j2], controlplane_instance_group.instances.all()[0])
+        TaskManager.start_task.assert_called_once_with(pu, controlplane_instance_group, controlplane_instance_group.instances.all()[0])
         pu.finished = pu.created + timedelta(seconds=1)
         pu.status = "successful"
         pu.save()
@@ -62,8 +62,8 @@ def test_multi_group_with_shared_dependency(instance_factory, controlplane_insta
         DependencyManager().schedule()
         TaskManager().schedule()
 
-        TaskManager.start_task.assert_any_call(j1, ig1, [], i1)
-        TaskManager.start_task.assert_any_call(j2, ig2, [], i2)
+        TaskManager.start_task.assert_any_call(j1, ig1, i1)
+        TaskManager.start_task.assert_any_call(j2, ig2, i2)
         assert TaskManager.start_task.call_count == 2
 
 
@@ -75,7 +75,7 @@ def test_workflow_job_no_instancegroup(workflow_job_template_factory, controlpla
     wfj.save()
     with mocker.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(wfj, None, [], None)
+        TaskManager.start_task.assert_called_once_with(wfj, None, None)
         assert wfj.instance_group is None
 
 
@@ -150,7 +150,7 @@ def test_failover_group_run(instance_factory, controlplane_instance_group, mocke
         mock_task_impact.return_value = 500
         with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
             tm.schedule()
-            mock_job.assert_has_calls([mock.call(j1, ig1, [], i1), mock.call(j1_1, ig2, [], i2)])
+            mock_job.assert_has_calls([mock.call(j1, ig1, i1), mock.call(j1_1, ig2, i2)])
             assert mock_job.call_count == 2
 
 

awx/main/tests/functional/task_management/test_scheduler.py

@@ -18,7 +18,7 @@ def test_single_job_scheduler_launch(hybrid_instance, controlplane_instance_grou
     j = create_job(objects.job_template)
     with mocker.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db
@@ -240,12 +240,82 @@ def test_multi_jt_capacity_blocking(hybrid_instance, job_template_factory, mocke
         mock_task_impact.return_value = 505
         with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
             tm.schedule()
-            mock_job.assert_called_once_with(j1, controlplane_instance_group, [], instance)
+            mock_job.assert_called_once_with(j1, controlplane_instance_group, instance)
             j1.status = "successful"
             j1.save()
     with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
         tm.schedule()
-        mock_job.assert_called_once_with(j2, controlplane_instance_group, [], instance)
+        mock_job.assert_called_once_with(j2, controlplane_instance_group, instance)
+
+
+@pytest.mark.django_db
+def test_max_concurrent_jobs_ig_capacity_blocking(hybrid_instance, job_template_factory, mocker):
+    """When max_concurrent_jobs of an instance group is more restrictive than capacity of instances, enforce max_concurrent_jobs."""
+    instance = hybrid_instance
+    controlplane_instance_group = instance.rampart_groups.first()
+    # We will expect only 1 job to be started
+    controlplane_instance_group.max_concurrent_jobs = 1
+    controlplane_instance_group.save()
+    num_jobs = 3
+    jobs = []
+    for i in range(num_jobs):
+        jobs.append(
+            create_job(job_template_factory(f'jt{i}', organization=f'org{i}', project=f'proj{i}', inventory=f'inv{i}', credential=f'cred{i}').job_template)
+        )
+    tm = TaskManager()
+    task_impact = 1
+
+    # Sanity check that multiple jobs would run if not for the max_concurrent_jobs setting.
+    assert task_impact * num_jobs < controlplane_instance_group.capacity
+    tm = TaskManager()
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = task_impact
+        with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
+            tm.schedule()
+            mock_job.assert_called_once()
+            jobs[0].status = 'running'
+            jobs[0].controller_node = instance.hostname
+            jobs[0].execution_node = instance.hostname
+            jobs[0].instance_group = controlplane_instance_group
+            jobs[0].save()
+
+    # while that job is running, we should not start another job
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = task_impact
+        with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
+            tm.schedule()
+            mock_job.assert_not_called()
+    # now job is done, we should start one of the two other jobs
+    jobs[0].status = 'successful'
+    jobs[0].save()
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = task_impact
+        with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
+            tm.schedule()
+            mock_job.assert_called_once()
+
+
+@pytest.mark.django_db
+def test_max_forks_ig_capacity_blocking(hybrid_instance, job_template_factory, mocker):
+    """When max_forks of an instance group is less than the capacity of instances, enforce max_forks."""
+    instance = hybrid_instance
+    controlplane_instance_group = instance.rampart_groups.first()
+    controlplane_instance_group.max_forks = 15
+    controlplane_instance_group.save()
+    task_impact = 10
+    num_jobs = 2
+    # Sanity check that 2 jobs would run if not for the max_forks setting.
+    assert controlplane_instance_group.max_forks < controlplane_instance_group.capacity
+    assert task_impact * num_jobs > controlplane_instance_group.max_forks
+    assert task_impact * num_jobs < controlplane_instance_group.capacity
+    for i in range(num_jobs):
+        create_job(job_template_factory(f'jt{i}', organization=f'org{i}', project=f'proj{i}', inventory=f'inv{i}', credential=f'cred{i}').job_template)
+    tm = TaskManager()
+    with mock.patch('awx.main.models.Job.task_impact', new_callable=mock.PropertyMock) as mock_task_impact:
+        mock_task_impact.return_value = task_impact
+        with mock.patch.object(TaskManager, "start_task", wraps=tm.start_task) as mock_job:
+            tm.schedule()
+            mock_job.assert_called_once()
 
 
 @pytest.mark.django_db
@@ -267,12 +337,12 @@ def test_single_job_dependencies_project_launch(controlplane_instance_group, job
             pu = [x for x in p.project_updates.all()]
             assert len(pu) == 1
             TaskManager().schedule()
-            TaskManager.start_task.assert_called_once_with(pu[0], controlplane_instance_group, [j], instance)
+            TaskManager.start_task.assert_called_once_with(pu[0], controlplane_instance_group, instance)
             pu[0].status = "successful"
             pu[0].save()
     with mock.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db
@@ -295,12 +365,12 @@ def test_single_job_dependencies_inventory_update_launch(controlplane_instance_g
             iu = [x for x in ii.inventory_updates.all()]
             assert len(iu) == 1
             TaskManager().schedule()
-            TaskManager.start_task.assert_called_once_with(iu[0], controlplane_instance_group, [j], instance)
+            TaskManager.start_task.assert_called_once_with(iu[0], controlplane_instance_group, instance)
             iu[0].status = "successful"
             iu[0].save()
     with mock.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db
@@ -342,7 +412,7 @@ def test_job_dependency_with_already_updated(controlplane_instance_group, job_te
             mock_iu.assert_not_called()
     with mock.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(j, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db
@@ -372,9 +442,7 @@ def test_shared_dependencies_launch(controlplane_instance_group, job_template_fa
         TaskManager().schedule()
         pu = p.project_updates.first()
         iu = ii.inventory_updates.first()
-        TaskManager.start_task.assert_has_calls(
-            [mock.call(iu, controlplane_instance_group, [j1, j2], instance), mock.call(pu, controlplane_instance_group, [j1, j2], instance)]
-        )
+        TaskManager.start_task.assert_has_calls([mock.call(iu, controlplane_instance_group, instance), mock.call(pu, controlplane_instance_group, instance)])
         pu.status = "successful"
         pu.finished = pu.created + timedelta(seconds=1)
         pu.save()
@@ -383,9 +451,7 @@ def test_shared_dependencies_launch(controlplane_instance_group, job_template_fa
         iu.save()
     with mock.patch("awx.main.scheduler.TaskManager.start_task"):
         TaskManager().schedule()
-        TaskManager.start_task.assert_has_calls(
-            [mock.call(j1, controlplane_instance_group, [], instance), mock.call(j2, controlplane_instance_group, [], instance)]
-        )
+        TaskManager.start_task.assert_has_calls([mock.call(j1, controlplane_instance_group, instance), mock.call(j2, controlplane_instance_group, instance)])
     pu = [x for x in p.project_updates.all()]
     iu = [x for x in ii.inventory_updates.all()]
     assert len(pu) == 1
@@ -409,7 +475,7 @@ def test_job_not_blocking_project_update(controlplane_instance_group, job_templa
         project_update.status = "pending"
         project_update.save()
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(project_update, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(project_update, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db
@@ -433,7 +499,7 @@ def test_job_not_blocking_inventory_update(controlplane_instance_group, job_temp
 
         DependencyManager().schedule()
         TaskManager().schedule()
-        TaskManager.start_task.assert_called_once_with(inventory_update, controlplane_instance_group, [], instance)
+        TaskManager.start_task.assert_called_once_with(inventory_update, controlplane_instance_group, instance)
 
 
 @pytest.mark.django_db

awx/main/tests/functional/test_instances.py

@@ -1,7 +1,7 @@
 import pytest
 from unittest import mock
 
-from awx.main.models import AdHocCommand, InventoryUpdate, JobTemplate
+from awx.main.models import AdHocCommand, InventoryUpdate, JobTemplate, Job
 from awx.main.models.activity_stream import ActivityStream
 from awx.main.models.ha import Instance, InstanceGroup
 from awx.main.tasks.system import apply_cluster_membership_policies
@@ -15,6 +15,24 @@ def test_default_tower_instance_group(default_instance_group, job_factory):
     assert default_instance_group in job_factory().preferred_instance_groups
 
 
+@pytest.mark.django_db
+@pytest.mark.parametrize('node_type', ('execution', 'control'))
+@pytest.mark.parametrize('active', (True, False))
+def test_get_cleanup_task_kwargs_active_jobs(node_type, active):
+    instance = Instance.objects.create(hostname='foobar', node_type=node_type)
+    job_kwargs = dict()
+    job_kwargs['controller_node' if node_type == 'control' else 'execution_node'] = instance.hostname
+    job_kwargs['status'] = 'running' if active else 'successful'
+
+    job = Job.objects.create(**job_kwargs)
+    kwargs = instance.get_cleanup_task_kwargs()
+
+    if active:
+        assert kwargs['exclude_strings'] == [f'awx_{job.pk}_']
+    else:
+        assert 'exclude_strings' not in kwargs
+
+
 @pytest.mark.django_db
 class TestPolicyTaskScheduling:
     """Tests make assertions about when the policy task gets scheduled"""

awx/main/tests/functional/test_licenses.py

@@ -121,8 +121,8 @@ def test_python_and_js_licenses():
         return errors
 
     base_dir = settings.BASE_DIR
-    api_licenses = index_licenses('%s/../docs/licenses' % base_dir)
-    ui_licenses = index_licenses('%s/../docs/licenses/ui' % base_dir)
+    api_licenses = index_licenses('%s/../licenses' % base_dir)
+    ui_licenses = index_licenses('%s/../licenses/ui' % base_dir)
     api_requirements = read_api_requirements('%s/../requirements' % base_dir)
     ui_requirements = read_ui_requirements('%s/ui' % base_dir)
 

awx/main/tests/functional/test_tasks.py

@@ -5,8 +5,8 @@ import tempfile
 import shutil
 
 from awx.main.tasks.jobs import RunJob
-from awx.main.tasks.system import execution_node_health_check, _cleanup_images_and_files
-from awx.main.models import Instance, Job
+from awx.main.tasks.system import execution_node_health_check, _cleanup_images_and_files, handle_work_error
+from awx.main.models import Instance, Job, InventoryUpdate, ProjectUpdate
 
 
 @pytest.fixture
@@ -74,3 +74,17 @@ def test_does_not_run_reaped_job(mocker, mock_me):
     job.refresh_from_db()
     assert job.status == 'failed'
     mock_run.assert_not_called()
+
+
+@pytest.mark.django_db
+def test_handle_work_error_nested(project, inventory_source):
+    pu = ProjectUpdate.objects.create(status='failed', project=project, celery_task_id='1234')
+    iu = InventoryUpdate.objects.create(status='pending', inventory_source=inventory_source, source='scm')
+    job = Job.objects.create(status='pending')
+    iu.dependent_jobs.add(pu)
+    job.dependent_jobs.add(pu, iu)
+    handle_work_error({'type': 'project_update', 'id': pu.id})
+    iu.refresh_from_db()
+    job.refresh_from_db()
+    assert iu.job_explanation == f'Previous Task Failed: {{"job_type": "project_update", "job_name": "", "job_id": "{pu.id}"}}'
+    assert job.job_explanation == f'Previous Task Failed: {{"job_type": "inventory_update", "job_name": "", "job_id": "{iu.id}"}}'

awx/main/tests/unit/models/test_ha.py

@@ -1,10 +1,7 @@
 import pytest
-from unittest import mock
-from unittest.mock import Mock
 from decimal import Decimal
 
-from awx.main.models import InstanceGroup, Instance
-from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups
+from awx.main.models import Instance
 
 
 @pytest.mark.parametrize('capacity_adjustment', [0.0, 0.25, 0.5, 0.75, 1, 1.5, 3])
@@ -17,83 +14,6 @@ def test_capacity_adjustment_no_save(capacity_adjustment):
     assert inst.capacity == (float(inst.capacity_adjustment) * abs(inst.mem_capacity - inst.cpu_capacity) + min(inst.mem_capacity, inst.cpu_capacity))
 
 
-def T(impact):
-    j = mock.Mock(spec_set=['task_impact', 'capacity_type'])
-    j.task_impact = impact
-    j.capacity_type = 'execution'
-    return j
-
-
-def Is(param):
-    """
-    param:
-        [remaining_capacity1, remaining_capacity2, remaining_capacity3, ...]
-        [(jobs_running1, capacity1), (jobs_running2, capacity2), (jobs_running3, capacity3), ...]
-    """
-
-    instances = []
-    if isinstance(param[0], tuple):
-        for (jobs_running, capacity) in param:
-            inst = Mock()
-            inst.capacity = capacity
-            inst.jobs_running = jobs_running
-            inst.node_type = 'execution'
-            instances.append(inst)
-    else:
-        for i in param:
-            inst = Mock()
-            inst.remaining_capacity = i
-            inst.node_type = 'execution'
-            instances.append(inst)
-    return instances
-
-
-class TestInstanceGroup(object):
-    @pytest.mark.parametrize(
-        'task,instances,instance_fit_index,reason',
-        [
-            (T(100), Is([100]), 0, "Only one, pick it"),
-            (T(100), Is([100, 100]), 0, "Two equally good fits, pick the first"),
-            (T(100), Is([50, 100]), 1, "First instance not as good as second instance"),
-            (T(100), Is([50, 0, 20, 100, 100, 100, 30, 20]), 3, "Pick Instance [3] as it is the first that the task fits in."),
-            (T(100), Is([50, 0, 20, 99, 11, 1, 5, 99]), None, "The task don't a fit, you must a quit!"),
-        ],
-    )
-    def test_fit_task_to_most_remaining_capacity_instance(self, task, instances, instance_fit_index, reason):
-        InstanceGroup(id=10)
-        tm_igs = TaskManagerInstanceGroups(instance_groups={'controlplane': {'instances': instances}})
-
-        instance_picked = tm_igs.fit_task_to_most_remaining_capacity_instance(task, 'controlplane')
-
-        if instance_fit_index is None:
-            assert instance_picked is None, reason
-        else:
-            assert instance_picked == instances[instance_fit_index], reason
-
-    @pytest.mark.parametrize(
-        'instances,instance_fit_index,reason',
-        [
-            (Is([(0, 100)]), 0, "One idle instance, pick it"),
-            (Is([(1, 100)]), None, "One un-idle instance, pick nothing"),
-            (Is([(0, 100), (0, 200), (1, 500), (0, 700)]), 3, "Pick the largest idle instance"),
-            (Is([(0, 100), (0, 200), (1, 10000), (0, 700), (0, 699)]), 3, "Pick the largest idle instance"),
-            (Is([(0, 0)]), None, "One idle but down instance, don't pick it"),
-        ],
-    )
-    def test_find_largest_idle_instance(self, instances, instance_fit_index, reason):
-        def filter_offline_instances(*args):
-            return filter(lambda i: i.capacity > 0, instances)
-
-        InstanceGroup(id=10)
-        instances_online_only = filter_offline_instances(instances)
-        tm_igs = TaskManagerInstanceGroups(instance_groups={'controlplane': {'instances': instances_online_only}})
-
-        if instance_fit_index is None:
-            assert tm_igs.find_largest_idle_instance('controlplane') is None, reason
-        else:
-            assert tm_igs.find_largest_idle_instance('controlplane') == instances[instance_fit_index], reason
-
-
 def test_cleanup_params_defaults():
     inst = Instance(hostname='foobar')
     assert inst.get_cleanup_task_kwargs(exclude_strings=['awx_423_']) == {'exclude_strings': ['awx_423_'], 'file_pattern': '/tmp/awx_*_*', 'grace_period': 60}

awx/main/tests/unit/models/test_jobs.py

@@ -36,15 +36,14 @@ def job(mocker, hosts, inventory):
 
 def test_start_job_fact_cache(hosts, job, inventory, tmpdir):
     fact_cache = os.path.join(tmpdir, 'facts')
-    modified_times = {}
-    job.start_job_fact_cache(fact_cache, modified_times, 0)
+    last_modified = job.start_job_fact_cache(fact_cache, timeout=0)
 
     for host in hosts:
         filepath = os.path.join(fact_cache, host.name)
         assert os.path.exists(filepath)
         with open(filepath, 'r') as f:
             assert f.read() == json.dumps(host.ansible_facts)
-        assert filepath in modified_times
+        assert os.path.getmtime(filepath) <= last_modified
 
 
 def test_fact_cache_with_invalid_path_traversal(job, inventory, tmpdir, mocker):
@@ -58,18 +57,16 @@ def test_fact_cache_with_invalid_path_traversal(job, inventory, tmpdir, mocker):
     )
 
     fact_cache = os.path.join(tmpdir, 'facts')
-    job.start_job_fact_cache(fact_cache, {}, 0)
+    job.start_job_fact_cache(fact_cache, timeout=0)
     # a file called "foo" should _not_ be written outside the facts dir
     assert os.listdir(os.path.join(fact_cache, '..')) == ['facts']
 
 
 def test_finish_job_fact_cache_with_existing_data(job, hosts, inventory, mocker, tmpdir):
     fact_cache = os.path.join(tmpdir, 'facts')
-    modified_times = {}
-    job.start_job_fact_cache(fact_cache, modified_times, 0)
+    last_modified = job.start_job_fact_cache(fact_cache, timeout=0)
 
-    for h in hosts:
-        h.save = mocker.Mock()
+    bulk_update = mocker.patch('django.db.models.query.QuerySet.bulk_update')
 
     ansible_facts_new = {"foo": "bar"}
     filepath = os.path.join(fact_cache, hosts[1].name)
@@ -83,23 +80,20 @@ def test_finish_job_fact_cache_with_existing_data(job, hosts, inventory, mocker,
         new_modification_time = time.time() + 3600
         os.utime(filepath, (new_modification_time, new_modification_time))
 
-    job.finish_job_fact_cache(fact_cache, modified_times)
+    job.finish_job_fact_cache(fact_cache, last_modified)
 
     for host in (hosts[0], hosts[2], hosts[3]):
-        host.save.assert_not_called()
         assert host.ansible_facts == {"a": 1, "b": 2}
         assert host.ansible_facts_modified is None
     assert hosts[1].ansible_facts == ansible_facts_new
-    hosts[1].save.assert_called_once_with(update_fields=['ansible_facts', 'ansible_facts_modified'])
+    bulk_update.assert_called_once_with([hosts[1]], ['ansible_facts', 'ansible_facts_modified'])
 
 
 def test_finish_job_fact_cache_with_bad_data(job, hosts, inventory, mocker, tmpdir):
     fact_cache = os.path.join(tmpdir, 'facts')
-    modified_times = {}
-    job.start_job_fact_cache(fact_cache, modified_times, 0)
+    last_modified = job.start_job_fact_cache(fact_cache, timeout=0)
 
-    for h in hosts:
-        h.save = mocker.Mock()
+    bulk_update = mocker.patch('django.db.models.query.QuerySet.bulk_update')
 
     for h in hosts:
         filepath = os.path.join(fact_cache, h.name)
@@ -109,26 +103,22 @@ def test_finish_job_fact_cache_with_bad_data(job, hosts, inventory, mocker, tmpd
             new_modification_time = time.time() + 3600
             os.utime(filepath, (new_modification_time, new_modification_time))
 
-    job.finish_job_fact_cache(fact_cache, modified_times)
+    job.finish_job_fact_cache(fact_cache, last_modified)
 
-    for h in hosts:
-        h.save.assert_not_called()
+    bulk_update.assert_not_called()
 
 
 def test_finish_job_fact_cache_clear(job, hosts, inventory, mocker, tmpdir):
     fact_cache = os.path.join(tmpdir, 'facts')
-    modified_times = {}
-    job.start_job_fact_cache(fact_cache, modified_times, 0)
+    last_modified = job.start_job_fact_cache(fact_cache, timeout=0)
 
-    for h in hosts:
-        h.save = mocker.Mock()
+    bulk_update = mocker.patch('django.db.models.query.QuerySet.bulk_update')
 
     os.remove(os.path.join(fact_cache, hosts[1].name))
-    job.finish_job_fact_cache(fact_cache, modified_times)
+    job.finish_job_fact_cache(fact_cache, last_modified)
 
     for host in (hosts[0], hosts[2], hosts[3]):
-        host.save.assert_not_called()
         assert host.ansible_facts == {"a": 1, "b": 2}
         assert host.ansible_facts_modified is None
     assert hosts[1].ansible_facts == {}
-    hosts[1].save.assert_called_once_with()
+    bulk_update.assert_called_once_with([hosts[1]], ['ansible_facts', 'ansible_facts_modified'])

awx/main/tests/unit/test_capacity.py

@@ -1,6 +1,6 @@
 import pytest
 
-from awx.main.scheduler.task_manager_models import TaskManagerInstanceGroups, TaskManagerInstances
+from awx.main.scheduler.task_manager_models import TaskManagerModels
 
 
 class FakeMeta(object):
@@ -16,38 +16,64 @@ class FakeObject(object):
 
 
 class Job(FakeObject):
-    task_impact = 43
-    is_container_group_task = False
-    controller_node = ''
-    execution_node = ''
+    def __init__(self, **kwargs):
+        self.task_impact = kwargs.get('task_impact', 43)
+        self.is_container_group_task = kwargs.get('is_container_group_task', False)
+        self.controller_node = kwargs.get('controller_node', '')
+        self.execution_node = kwargs.get('execution_node', '')
+        self.instance_group = kwargs.get('instance_group', None)
+        self.instance_group_id = self.instance_group.id if self.instance_group else None
+        self.capacity_type = kwargs.get('capacity_type', 'execution')
 
     def log_format(self):
         return 'job 382 (fake)'
 
 
+class Instances(FakeObject):
+    def add(self, *args):
+        for instance in args:
+            self.obj.instance_list.append(instance)
+
+    def all(self):
+        return self.obj.instance_list
+
+
+class InstanceGroup(FakeObject):
+    def __init__(self, **kwargs):
+        super(InstanceGroup, self).__init__(**kwargs)
+        self.instance_list = []
+        self.pk = self.id = kwargs.get('id', 1)
+
+    @property
+    def instances(self):
+        mgr = Instances(obj=self)
+        return mgr
+
+    @property
+    def is_container_group(self):
+        return False
+
+    @property
+    def max_concurrent_jobs(self):
+        return 0
+
+    @property
+    def max_forks(self):
+        return 0
+
+
+class Instance(FakeObject):
+    def __init__(self, **kwargs):
+        self.node_type = kwargs.get('node_type', 'hybrid')
+        self.capacity = kwargs.get('capacity', 0)
+        self.hostname = kwargs.get('hostname', 'fakehostname')
+        self.consumed_capacity = 0
+        self.jobs_running = 0
+
+
 @pytest.fixture
 def sample_cluster():
     def stand_up_cluster():
-        class Instances(FakeObject):
-            def add(self, *args):
-                for instance in args:
-                    self.obj.instance_list.append(instance)
-
-            def all(self):
-                return self.obj.instance_list
-
-        class InstanceGroup(FakeObject):
-            def __init__(self, **kwargs):
-                super(InstanceGroup, self).__init__(**kwargs)
-                self.instance_list = []
-
-            @property
-            def instances(self):
-                mgr = Instances(obj=self)
-                return mgr
-
-        class Instance(FakeObject):
-            pass
 
         ig_small = InstanceGroup(name='ig_small')
         ig_large = InstanceGroup(name='ig_large')
@@ -66,14 +92,12 @@ def sample_cluster():
 @pytest.fixture
 def create_ig_manager():
     def _rf(ig_list, tasks):
-        instances = TaskManagerInstances(tasks, instances=set(inst for ig in ig_list for inst in ig.instance_list))
-
-        seed_igs = {}
-        for ig in ig_list:
-            seed_igs[ig.name] = {'instances': [instances[inst.hostname] for inst in ig.instance_list]}
-
-        instance_groups = TaskManagerInstanceGroups(instance_groups=seed_igs)
-        return instance_groups
+        tm_models = TaskManagerModels.init_with_consumed_capacity(
+            tasks=tasks,
+            instances=set(inst for ig in ig_list for inst in ig.instance_list),
+            instance_groups=ig_list,
+        )
+        return tm_models.instance_groups
 
     return _rf
 
@@ -126,3 +150,75 @@ def test_RBAC_reduced_filter(sample_cluster, create_ig_manager):
     # Cross-links between groups not visible to current user,
     # so a naieve accounting of capacities is returned instead
     assert instance_groups_mgr.get_consumed_capacity('default') == 43
+
+
+def Is(param):
+    """
+    param:
+        [remaining_capacity1, remaining_capacity2, remaining_capacity3, ...]
+        [(jobs_running1, capacity1), (jobs_running2, capacity2), (jobs_running3, capacity3), ...]
+    """
+
+    instances = []
+    if isinstance(param[0], tuple):
+        for index, (jobs_running, capacity) in enumerate(param):
+            inst = Instance(capacity=capacity, node_type='execution', hostname=f'fakehost-{index}')
+            inst.jobs_running = jobs_running
+            instances.append(inst)
+    else:
+        for index, capacity in enumerate(param):
+            inst = Instance(capacity=capacity, node_type='execution', hostname=f'fakehost-{index}')
+            inst.node_type = 'execution'
+            instances.append(inst)
+    return instances
+
+
+class TestSelectBestInstanceForTask(object):
+    @pytest.mark.parametrize(
+        'task,instances,instance_fit_index,reason',
+        [
+            (Job(task_impact=100), Is([100]), 0, "Only one, pick it"),
+            (Job(task_impact=100), Is([100, 100]), 0, "Two equally good fits, pick the first"),
+            (Job(task_impact=100), Is([50, 100]), 1, "First instance not as good as second instance"),
+            (Job(task_impact=100), Is([50, 0, 20, 100, 100, 100, 30, 20]), 3, "Pick Instance [3] as it is the first that the task fits in."),
+            (Job(task_impact=100), Is([50, 0, 20, 99, 11, 1, 5, 99]), None, "The task don't a fit, you must a quit!"),
+        ],
+    )
+    def test_fit_task_to_most_remaining_capacity_instance(self, task, instances, instance_fit_index, reason):
+        ig = InstanceGroup(id=10, name='controlplane')
+        tasks = []
+        for instance in instances:
+            ig.instances.add(instance)
+            for _ in range(instance.jobs_running):
+                tasks.append(Job(execution_node=instance.hostname, controller_node=instance.hostname, instance_group=ig))
+        tm_models = TaskManagerModels.init_with_consumed_capacity(tasks=tasks, instances=instances, instance_groups=[ig])
+        instance_picked = tm_models.instance_groups.fit_task_to_most_remaining_capacity_instance(task, 'controlplane')
+
+        if instance_fit_index is None:
+            assert instance_picked is None, reason
+        else:
+            assert instance_picked.hostname == instances[instance_fit_index].hostname, reason
+
+    @pytest.mark.parametrize(
+        'instances,instance_fit_index,reason',
+        [
+            (Is([(0, 100)]), 0, "One idle instance, pick it"),
+            (Is([(1, 100)]), None, "One un-idle instance, pick nothing"),
+            (Is([(0, 100), (0, 200), (1, 500), (0, 700)]), 3, "Pick the largest idle instance"),
+            (Is([(0, 100), (0, 200), (1, 10000), (0, 700), (0, 699)]), 3, "Pick the largest idle instance"),
+            (Is([(0, 0)]), None, "One idle but down instance, don't pick it"),
+        ],
+    )
+    def test_find_largest_idle_instance(self, instances, instance_fit_index, reason):
+        ig = InstanceGroup(id=10, name='controlplane')
+        tasks = []
+        for instance in instances:
+            ig.instances.add(instance)
+            for _ in range(instance.jobs_running):
+                tasks.append(Job(execution_node=instance.hostname, controller_node=instance.hostname, instance_group=ig))
+        tm_models = TaskManagerModels.init_with_consumed_capacity(tasks=tasks, instances=instances, instance_groups=[ig])
+
+        if instance_fit_index is None:
+            assert tm_models.instance_groups.find_largest_idle_instance('controlplane') is None, reason
+        else:
+            assert tm_models.instance_groups.find_largest_idle_instance('controlplane').hostname == instances[instance_fit_index].hostname, reason

awx/main/utils/common.py

@@ -11,11 +11,12 @@ import os
 import subprocess
 import re
 import stat
+import sys
 import urllib.parse
 import threading
 import contextlib
 import tempfile
-from functools import reduce, wraps
+import functools
 
 # Django
 from django.core.exceptions import ObjectDoesNotExist, FieldDoesNotExist
@@ -73,6 +74,7 @@ __all__ = [
     'NullablePromptPseudoField',
     'model_instance_diff',
     'parse_yaml_or_json',
+    'is_testing',
     'RequireDebugTrueOrTest',
     'has_model_field_prefetched',
     'set_environ',
@@ -88,6 +90,7 @@ __all__ = [
     'deepmerge',
     'get_event_partition_epoch',
     'cleanup_new_process',
+    'log_excess_runtime',
 ]
 
 
@@ -144,6 +147,19 @@ def underscore_to_camelcase(s):
     return ''.join(x.capitalize() or '_' for x in s.split('_'))
 
 
+@functools.cache
+def is_testing(argv=None):
+    '''Return True if running django or py.test unit tests.'''
+    if 'PYTEST_CURRENT_TEST' in os.environ.keys():
+        return True
+    argv = sys.argv if argv is None else argv
+    if len(argv) >= 1 and ('py.test' in argv[0] or 'py/test.py' in argv[0]):
+        return True
+    elif len(argv) >= 2 and argv[1] == 'test':
+        return True
+    return False
+
+
 class RequireDebugTrueOrTest(logging.Filter):
     """
     Logging filter to output when in DEBUG mode or running tests.
@@ -152,7 +168,7 @@ class RequireDebugTrueOrTest(logging.Filter):
     def filter(self, record):
         from django.conf import settings
 
-        return settings.DEBUG or settings.IS_TESTING()
+        return settings.DEBUG or is_testing()
 
 
 class IllegalArgumentError(ValueError):
@@ -174,7 +190,7 @@ def memoize(ttl=60, cache_key=None, track_function=False, cache=None):
     cache = cache or get_memoize_cache()
 
     def memoize_decorator(f):
-        @wraps(f)
+        @functools.wraps(f)
         def _memoizer(*args, **kwargs):
             if track_function:
                 cache_dict_key = slugify('%r %r' % (args, kwargs))
@@ -992,7 +1008,7 @@ def getattrd(obj, name, default=NoDefaultProvided):
     """
 
     try:
-        return reduce(getattr, name.split("."), obj)
+        return functools.reduce(getattr, name.split("."), obj)
     except AttributeError:
         if default != NoDefaultProvided:
             return default
@@ -1188,7 +1204,7 @@ def cleanup_new_process(func):
     Cleanup django connection, cache connection, before executing new thread or processes entry point, func.
     """
 
-    @wraps(func)
+    @functools.wraps(func)
     def wrapper_cleanup_new_process(*args, **kwargs):
         from awx.conf.settings import SettingsWrapper  # noqa
 
@@ -1200,15 +1216,30 @@ def cleanup_new_process(func):
     return wrapper_cleanup_new_process
 
 
-def log_excess_runtime(func_logger, cutoff=5.0):
+def log_excess_runtime(func_logger, cutoff=5.0, debug_cutoff=5.0, msg=None, add_log_data=False):
     def log_excess_runtime_decorator(func):
-        @wraps(func)
+        @functools.wraps(func)
         def _new_func(*args, **kwargs):
             start_time = time.time()
-            return_value = func(*args, **kwargs)
-            delta = time.time() - start_time
-            if delta > cutoff:
-                logger.info(f'Running {func.__name__!r} took {delta:.2f}s')
+            log_data = {'name': repr(func.__name__)}
+
+            if add_log_data:
+                return_value = func(*args, log_data=log_data, **kwargs)
+            else:
+                return_value = func(*args, **kwargs)
+
+            log_data['delta'] = time.time() - start_time
+            if isinstance(return_value, dict):
+                log_data.update(return_value)
+
+            if msg is None:
+                record_msg = 'Running {name} took {delta:.2f}s'
+            else:
+                record_msg = msg
+            if log_data['delta'] > cutoff:
+                func_logger.info(record_msg.format(**log_data))
+            elif log_data['delta'] > debug_cutoff:
+                func_logger.debug(record_msg.format(**log_data))
             return return_value
 
         return _new_func

awx/main/wsbroadcast.py

@@ -118,7 +118,7 @@ class WebsocketTask:
             logger.warning(f"Connection from {self.name} to {self.remote_host} timed out.")
         except Exception as e:
             # Early on, this is our canary. I'm not sure what exceptions we can really encounter.
-            logger.warning(f"Connection from {self.name} to {self.remote_host} failed for unknown reason: '{e}'.")
+            logger.exception(f"Connection from {self.name} to {self.remote_host} failed for unknown reason: '{e}'.")
         else:
             logger.warning(f"Connection from {self.name} to {self.remote_host} list.")
 

awx/settings/defaults.py

@@ -10,28 +10,6 @@ import socket
 from datetime import timedelta
 
 
-# Build paths inside the project like this: os.path.join(BASE_DIR, ...)
-BASE_DIR = os.path.dirname(os.path.dirname(__file__))
-
-
-def is_testing(argv=None):
-    import sys
-
-    '''Return True if running django or py.test unit tests.'''
-    if 'PYTEST_CURRENT_TEST' in os.environ.keys():
-        return True
-    argv = sys.argv if argv is None else argv
-    if len(argv) >= 1 and ('py.test' in argv[0] or 'py/test.py' in argv[0]):
-        return True
-    elif len(argv) >= 2 and argv[1] == 'test':
-        return True
-    return False
-
-
-def IS_TESTING(argv=None):
-    return is_testing(argv)
-
-
 if "pytest" in sys.modules:
     from unittest import mock
 
@@ -40,9 +18,13 @@ if "pytest" in sys.modules:
 else:
     import ldap
 
+
 DEBUG = True
 SQL_DEBUG = DEBUG
 
+# Build paths inside the project like this: os.path.join(BASE_DIR, ...)
+BASE_DIR = os.path.dirname(os.path.dirname(__file__))
+
 # FIXME: it would be nice to cycle back around and allow this to be
 # BigAutoField going forward, but we'd have to be explicit about our
 # existing models.
@@ -322,11 +304,13 @@ INSTALLED_APPS = [
     'django.contrib.messages',
     'django.contrib.sessions',
     'django.contrib.sites',
+    # daphne has to be installed before django.contrib.staticfiles for the app to startup
+    # According to channels 4.0 docs you install daphne instead of channels now
+    'daphne',
     'django.contrib.staticfiles',
     'oauth2_provider',
     'rest_framework',
     'django_extensions',
-    'channels',
     'polymorphic',
     'taggit',
     'social_django',
@@ -869,6 +853,7 @@ LOGGING = {
         'awx.main.signals': {'level': 'INFO'},  # very verbose debug-level logs
         'awx.api.permissions': {'level': 'INFO'},  # very verbose debug-level logs
         'awx.analytics': {'handlers': ['external_logger'], 'level': 'INFO', 'propagate': False},
+        'awx.analytics.broadcast_websocket': {'handlers': ['console', 'file', 'wsbroadcast', 'external_logger'], 'level': 'INFO', 'propagate': False},
         'awx.analytics.performance': {'handlers': ['console', 'file', 'tower_warnings', 'external_logger'], 'level': 'DEBUG', 'propagate': False},
         'awx.analytics.job_lifecycle': {'handlers': ['console', 'job_lifecycle'], 'level': 'DEBUG', 'propagate': False},
         'django_auth_ldap': {'handlers': ['console', 'file', 'tower_warnings'], 'level': 'DEBUG'},
@@ -1001,6 +986,13 @@ DJANGO_GUID = {'GUID_HEADER_NAME': 'X-API-Request-Id'}
 DEFAULT_EXECUTION_QUEUE_NAME = 'default'
 # pod spec used when the default execution queue is a container group, e.g. when deploying on k8s/ocp with the operator
 DEFAULT_EXECUTION_QUEUE_POD_SPEC_OVERRIDE = ''
+# Max number of concurrently consumed forks for the default execution queue
+# Zero means no limit
+DEFAULT_EXECUTION_QUEUE_MAX_FORKS = 0
+# Max number of concurrently running jobs for the default execution queue
+# Zero means no limit
+DEFAULT_EXECUTION_QUEUE_MAX_CONCURRENT_JOBS = 0
+
 # Name of the default controlplane queue
 DEFAULT_CONTROL_PLANE_QUEUE_NAME = 'controlplane'
 

awx/settings/development.py

@@ -114,7 +114,7 @@ if 'sqlite3' not in DATABASES['default']['ENGINE']:  # noqa
 # this needs to stay at the bottom of this file
 try:
     if os.getenv('AWX_KUBE_DEVEL', False):
-        include(optional('minikube.py'), scope=locals())
+        include(optional('development_kube.py'), scope=locals())
     else:
         include(optional('local_*.py'), scope=locals())
 except ImportError:

awx/settings/development_kube.py

@@ -1,4 +1,4 @@
 BROADCAST_WEBSOCKET_SECRET = '🤖starscream🤖'
-BROADCAST_WEBSOCKET_PORT = 8013
+BROADCAST_WEBSOCKET_PORT = 8052
 BROADCAST_WEBSOCKET_VERIFY_CERT = False
 BROADCAST_WEBSOCKET_PROTOCOL = 'http'

awx/ui/package-lock.json

@@ -8,7 +8,7 @@
       "dependencies": {
         "@lingui/react": "3.14.0",
         "@patternfly/patternfly": "4.217.1",
-        "@patternfly/react-core": "^4.250.1",
+        "@patternfly/react-core": "^4.264.0",
         "@patternfly/react-icons": "4.92.10",
         "@patternfly/react-table": "4.108.0",
         "ace-builds": "^1.10.1",
@@ -22,7 +22,7 @@
         "has-ansi": "5.0.1",
         "html-entities": "2.3.2",
         "js-yaml": "4.1.0",
-        "luxon": "^3.0.3",
+        "luxon": "^3.1.1",
         "prop-types": "^15.8.1",
         "react": "17.0.2",
         "react-ace": "^10.1.0",
@@ -3752,13 +3752,13 @@
       "integrity": "sha512-uN7JgfQsyR16YHkuGRCTIcBcnyKIqKjGkB2SGk9x1XXH3yYGenL83kpAavX9Xtozqp17KppOlybJuzcKvZMrgw=="
     },
     "node_modules/@patternfly/react-core": {
-      "version": "4.250.1",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-core/-/react-core-4.250.1.tgz",
-      "integrity": "sha512-vAOZPQdZzYXl/vkHnHMIt1eC3nrPDdsuuErPatkNPwmSvilXuXmWP5wxoJ36FbSNRRURkprFwx52zMmWS3iHJA==",
+      "version": "4.264.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-core/-/react-core-4.264.0.tgz",
+      "integrity": "sha512-tK0BMWxw8nhukev40HZ6q6d02pDnjX7oyA91vHa18aakJUKBWMaerqpG4NZVMoh0tPKX3aLNj+zyCwDALFAZZw==",
       "dependencies": {
-        "@patternfly/react-icons": "^4.92.6",
-        "@patternfly/react-styles": "^4.91.6",
-        "@patternfly/react-tokens": "^4.93.6",
+        "@patternfly/react-icons": "^4.93.0",
+        "@patternfly/react-styles": "^4.92.0",
+        "@patternfly/react-tokens": "^4.94.0",
         "focus-trap": "6.9.2",
         "react-dropzone": "9.0.0",
         "tippy.js": "5.1.2",
@@ -3769,6 +3769,15 @@
         "react-dom": "^16.8 || ^17 || ^18"
       }
     },
+    "node_modules/@patternfly/react-core/node_modules/@patternfly/react-icons": {
+      "version": "4.93.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-icons/-/react-icons-4.93.0.tgz",
+      "integrity": "sha512-OH0vORVioL+HLWMEog8/3u8jsiMCeJ0pFpvRKRhy5Uk4CdAe40k1SOBvXJP6opr+O8TLbz0q3bm8Jsh/bPaCuQ==",
+      "peerDependencies": {
+        "react": "^16.8 || ^17 || ^18",
+        "react-dom": "^16.8 || ^17 || ^18"
+      }
+    },
     "node_modules/@patternfly/react-core/node_modules/tslib": {
       "version": "2.3.1",
       "resolved": "https://registry.npmjs.org/tslib/-/tslib-2.3.1.tgz",
@@ -3784,9 +3793,9 @@
       }
     },
     "node_modules/@patternfly/react-styles": {
-      "version": "4.91.10",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-styles/-/react-styles-4.91.10.tgz",
-      "integrity": "sha512-fAG4Vjp63ohiR92F4e/Gkw5q1DSSckHKqdnEF75KUpSSBORzYP0EKMpupSd6ItpQFJw3iWs3MJi3/KIAAfU1Jw=="
+      "version": "4.92.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-styles/-/react-styles-4.92.0.tgz",
+      "integrity": "sha512-B/f6iyu8UEN1+wRxdC4sLIhvJeyL8SqInDXZmwOIqK8uPJ8Lze7qrbVhkkVzbMF37/oDPVa6dZH8qZFq062LEA=="
     },
     "node_modules/@patternfly/react-table": {
       "version": "4.108.0",
@@ -3811,9 +3820,9 @@
       "integrity": "sha512-d6xOpEDfsi2CZVlPQzGeux8XMwLT9hssAsaPYExaQMuYskwb+x1x7J371tWlbBdWHroy99KnVB6qIkUbs5X3UQ=="
     },
     "node_modules/@patternfly/react-tokens": {
-      "version": "4.93.10",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-tokens/-/react-tokens-4.93.10.tgz",
-      "integrity": "sha512-F+j1irDc9M6zvY6qNtDryhbpnHz3R8ymHRdGelNHQzPTIK88YSWEnT1c9iUI+uM/iuZol7sJmO5STtg2aPIDRQ=="
+      "version": "4.94.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-tokens/-/react-tokens-4.94.0.tgz",
+      "integrity": "sha512-fYXxUJZnzpn89K2zzHF0cSncZZVGKrohdb5f5T1wzxwU2NZPVGpvr88xhm+V2Y/fSrrTPwXcP3IIdtNOOtJdZw=="
     },
     "node_modules/@pmmmwh/react-refresh-webpack-plugin": {
       "version": "0.5.4",
@@ -15468,9 +15477,9 @@
       }
     },
     "node_modules/luxon": {
-      "version": "3.0.3",
-      "resolved": "https://registry.npmjs.org/luxon/-/luxon-3.0.3.tgz",
-      "integrity": "sha512-+EfHWnF+UT7GgTnq5zXg3ldnTKL2zdv7QJgsU5bjjpbH17E3qi/puMhQyJVYuCq+FRkogvB5WB6iVvUr+E4a7w==",
+      "version": "3.1.1",
+      "resolved": "https://registry.npmjs.org/luxon/-/luxon-3.1.1.tgz",
+      "integrity": "sha512-Ah6DloGmvseB/pX1cAmjbFvyU/pKuwQMQqz7d0yvuDlVYLTs2WeDHQMpC8tGjm1da+BriHROW/OEIT/KfYg6xw==",
       "engines": {
         "node": ">=12"
       }
@@ -25094,19 +25103,25 @@
       "integrity": "sha512-uN7JgfQsyR16YHkuGRCTIcBcnyKIqKjGkB2SGk9x1XXH3yYGenL83kpAavX9Xtozqp17KppOlybJuzcKvZMrgw=="
     },
     "@patternfly/react-core": {
-      "version": "4.250.1",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-core/-/react-core-4.250.1.tgz",
-      "integrity": "sha512-vAOZPQdZzYXl/vkHnHMIt1eC3nrPDdsuuErPatkNPwmSvilXuXmWP5wxoJ36FbSNRRURkprFwx52zMmWS3iHJA==",
+      "version": "4.264.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-core/-/react-core-4.264.0.tgz",
+      "integrity": "sha512-tK0BMWxw8nhukev40HZ6q6d02pDnjX7oyA91vHa18aakJUKBWMaerqpG4NZVMoh0tPKX3aLNj+zyCwDALFAZZw==",
       "requires": {
-        "@patternfly/react-icons": "^4.92.6",
-        "@patternfly/react-styles": "^4.91.6",
-        "@patternfly/react-tokens": "^4.93.6",
+        "@patternfly/react-icons": "^4.93.0",
+        "@patternfly/react-styles": "^4.92.0",
+        "@patternfly/react-tokens": "^4.94.0",
         "focus-trap": "6.9.2",
         "react-dropzone": "9.0.0",
         "tippy.js": "5.1.2",
         "tslib": "^2.0.0"
       },
       "dependencies": {
+        "@patternfly/react-icons": {
+          "version": "4.93.0",
+          "resolved": "https://registry.npmjs.org/@patternfly/react-icons/-/react-icons-4.93.0.tgz",
+          "integrity": "sha512-OH0vORVioL+HLWMEog8/3u8jsiMCeJ0pFpvRKRhy5Uk4CdAe40k1SOBvXJP6opr+O8TLbz0q3bm8Jsh/bPaCuQ==",
+          "requires": {}
+        },
         "tslib": {
           "version": "2.3.1",
           "resolved": "https://registry.npmjs.org/tslib/-/tslib-2.3.1.tgz",
@@ -25121,9 +25136,9 @@
       "requires": {}
     },
     "@patternfly/react-styles": {
-      "version": "4.91.10",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-styles/-/react-styles-4.91.10.tgz",
-      "integrity": "sha512-fAG4Vjp63ohiR92F4e/Gkw5q1DSSckHKqdnEF75KUpSSBORzYP0EKMpupSd6ItpQFJw3iWs3MJi3/KIAAfU1Jw=="
+      "version": "4.92.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-styles/-/react-styles-4.92.0.tgz",
+      "integrity": "sha512-B/f6iyu8UEN1+wRxdC4sLIhvJeyL8SqInDXZmwOIqK8uPJ8Lze7qrbVhkkVzbMF37/oDPVa6dZH8qZFq062LEA=="
     },
     "@patternfly/react-table": {
       "version": "4.108.0",
@@ -25146,9 +25161,9 @@
       }
     },
     "@patternfly/react-tokens": {
-      "version": "4.93.10",
-      "resolved": "https://registry.npmjs.org/@patternfly/react-tokens/-/react-tokens-4.93.10.tgz",
-      "integrity": "sha512-F+j1irDc9M6zvY6qNtDryhbpnHz3R8ymHRdGelNHQzPTIK88YSWEnT1c9iUI+uM/iuZol7sJmO5STtg2aPIDRQ=="
+      "version": "4.94.0",
+      "resolved": "https://registry.npmjs.org/@patternfly/react-tokens/-/react-tokens-4.94.0.tgz",
+      "integrity": "sha512-fYXxUJZnzpn89K2zzHF0cSncZZVGKrohdb5f5T1wzxwU2NZPVGpvr88xhm+V2Y/fSrrTPwXcP3IIdtNOOtJdZw=="
     },
     "@pmmmwh/react-refresh-webpack-plugin": {
       "version": "0.5.4",
@@ -34210,9 +34225,9 @@
       }
     },
     "luxon": {
-      "version": "3.0.3",
-      "resolved": "https://registry.npmjs.org/luxon/-/luxon-3.0.3.tgz",
-      "integrity": "sha512-+EfHWnF+UT7GgTnq5zXg3ldnTKL2zdv7QJgsU5bjjpbH17E3qi/puMhQyJVYuCq+FRkogvB5WB6iVvUr+E4a7w=="
+      "version": "3.1.1",
+      "resolved": "https://registry.npmjs.org/luxon/-/luxon-3.1.1.tgz",
+      "integrity": "sha512-Ah6DloGmvseB/pX1cAmjbFvyU/pKuwQMQqz7d0yvuDlVYLTs2WeDHQMpC8tGjm1da+BriHROW/OEIT/KfYg6xw=="
     },
     "lz-string": {
       "version": "1.4.4",

awx/ui/package.json

@@ -8,7 +8,7 @@
   "dependencies": {
     "@lingui/react": "3.14.0",
     "@patternfly/patternfly": "4.217.1",
-    "@patternfly/react-core": "^4.250.1",
+    "@patternfly/react-core": "^4.264.0",
     "@patternfly/react-icons": "4.92.10",
     "@patternfly/react-table": "4.108.0",
     "ace-builds": "^1.10.1",
@@ -22,7 +22,7 @@
     "has-ansi": "5.0.1",
     "html-entities": "2.3.2",
     "js-yaml": "4.1.0",
-    "luxon": "^3.0.3",
+    "luxon": "^3.1.1",
     "prop-types": "^15.8.1",
     "react": "17.0.2",
     "react-ace": "^10.1.0",

awx/ui/src/api/models/Hosts.js

@@ -20,6 +20,10 @@ class Hosts extends Base {
     return this.http.get(`${this.baseUrl}${id}/all_groups/`, { params });
   }
 
+  readGroups(id, params) {
+    return this.http.get(`${this.baseUrl}${id}/groups/`, { params });
+  }
+
   readGroupsOptions(id) {
     return this.http.options(`${this.baseUrl}${id}/groups/`);
   }

awx/ui/src/components/InstanceGroupLabels/InstanceGroupLabels.js

@@ -0,0 +1,49 @@
+import React from 'react';
+import { arrayOf, bool, number, shape, string } from 'prop-types';
+
+import { Label, LabelGroup } from '@patternfly/react-core';
+import { Link } from 'react-router-dom';
+
+function InstanceGroupLabels({ labels, isLinkable }) {
+  const buildLinkURL = (isContainerGroup) =>
+    isContainerGroup
+      ? '/instance_groups/container_group/'
+      : '/instance_groups/';
+  return (
+    <LabelGroup numLabels={5}>
+      {labels.map(({ id, name, is_container_group }) =>
+        isLinkable ? (
+          <Label
+            color="blue"
+            key={id}
+            render={({ className, content, componentRef }) => (
+              <Link
+                className={className}
+                innerRef={componentRef}
+                to={`${buildLinkURL(is_container_group)}${id}/details`}
+              >
+                {content}
+              </Link>
+            )}
+          >
+            {name}
+          </Label>
+        ) : (
+          <Label color="blue" key={id}>
+            {name}
+          </Label>
+        )
+      )}
+    </LabelGroup>
+  );
+}
+
+InstanceGroupLabels.propTypes = {
+  labels: arrayOf(shape({ id: number.isRequired, name: string.isRequired }))
+    .isRequired,
+  isLinkable: bool,
+};
+
+InstanceGroupLabels.defaultProps = { isLinkable: false };
+
+export default InstanceGroupLabels;

awx/ui/src/components/InstanceGroupLabels/index.js

@@ -0,0 +1 @@
+export { default } from './InstanceGroupLabels';

awx/ui/src/components/LaunchPrompt/steps/CredentialsStep.js

@@ -153,6 +153,10 @@ function CredentialsStep({
             }))}
             value={selectedType && selectedType.id}
             onChange={(e, id) => {
+              // Reset query params when the category of credentials is changed
+              history.replace({
+                search: '',
+              });
               setSelectedType(types.find((o) => o.id === parseInt(id, 10)));
             }}
           />

awx/ui/src/components/LaunchPrompt/steps/CredentialsStep.test.js

@@ -3,6 +3,7 @@ import { act } from 'react-dom/test-utils';
 import { Formik } from 'formik';
 import { CredentialsAPI, CredentialTypesAPI } from 'api';
 import { mountWithContexts } from '../../../../testUtils/enzymeHelpers';
+import { createMemoryHistory } from 'history';
 import CredentialsStep from './CredentialsStep';
 
 jest.mock('../../../api/models/CredentialTypes');
@@ -164,6 +165,41 @@ describe('CredentialsStep', () => {
     });
   });
 
+  test('should reset query params (credential.page) when selected credential type is changed', async () => {
+    let wrapper;
+    const history = createMemoryHistory({
+      initialEntries: ['?credential.page=2'],
+    });
+    await act(async () => {
+      wrapper = mountWithContexts(
+        <Formik>
+          <CredentialsStep allowCredentialsWithPasswords />
+        </Formik>,
+        {
+          context: { router: { history } },
+        }
+      );
+    });
+    wrapper.update();
+
+    expect(CredentialsAPI.read).toHaveBeenCalledWith({
+      credential_type: 1,
+      order_by: 'name',
+      page: 2,
+      page_size: 5,
+    });
+
+    await act(async () => {
+      wrapper.find('AnsibleSelect').invoke('onChange')({}, 3);
+    });
+    expect(CredentialsAPI.read).toHaveBeenCalledWith({
+      credential_type: 3,
+      order_by: 'name',
+      page: 1,
+      page_size: 5,
+    });
+  });
+
   test("error should be shown when a credential that prompts for passwords is selected on a step that doesn't allow it", async () => {
     let wrapper;
     await act(async () => {

awx/ui/src/components/Lookup/MultiCredentialsLookup.js

@@ -173,6 +173,10 @@ function MultiCredentialsLookup({
                 }))}
                 value={selectedType && selectedType.id}
                 onChange={(e, id) => {
+                  // Reset query params when the category of credentials is changed
+                  history.replace({
+                    search: '',
+                  });
                   setSelectedType(
                     credentialTypes.find((o) => o.id === parseInt(id, 10))
                   );

awx/ui/src/components/Lookup/MultiCredentialsLookup.test.js

@@ -6,6 +6,7 @@ import {
   mountWithContexts,
   waitForElement,
 } from '../../../testUtils/enzymeHelpers';
+import { createMemoryHistory } from 'history';
 import MultiCredentialsLookup from './MultiCredentialsLookup';
 
 jest.mock('../../api');
@@ -228,6 +229,53 @@ describe('<Formik><MultiCredentialsLookup /></Formik>', () => {
     ]);
   });
 
+  test('should reset query params (credentials.page) when selected credential type is changed', async () => {
+    const history = createMemoryHistory({
+      initialEntries: ['?credentials.page=2'],
+    });
+    await act(async () => {
+      wrapper = mountWithContexts(
+        <Formik>
+          <MultiCredentialsLookup
+            value={credentials}
+            tooltip="This is credentials look up"
+            onChange={() => {}}
+            onError={() => {}}
+          />
+        </Formik>,
+        {
+          context: { router: { history } },
+        }
+      );
+    });
+    const searchButton = await waitForElement(
+      wrapper,
+      'Button[aria-label="Search"]'
+    );
+    await act(async () => {
+      searchButton.invoke('onClick')();
+    });
+    expect(CredentialsAPI.read).toHaveBeenCalledWith({
+      credential_type: 400,
+      order_by: 'name',
+      page: 2,
+      page_size: 5,
+    });
+
+    const select = await waitForElement(wrapper, 'AnsibleSelect');
+    await act(async () => {
+      select.invoke('onChange')({}, 500);
+    });
+    wrapper.update();
+
+    expect(CredentialsAPI.read).toHaveBeenCalledWith({
+      credential_type: 500,
+      order_by: 'name',
+      page: 1,
+      page_size: 5,
+    });
+  });
+
   test('should only add 1 credential per credential type except vault(see below)', async () => {
     const onChange = jest.fn();
     await act(async () => {

awx/ui/src/components/PromptDetail/PromptDetail.js

@@ -6,6 +6,7 @@ import { Link } from 'react-router-dom';
 import styled from 'styled-components';
 import { Chip, Divider, Title } from '@patternfly/react-core';
 import { toTitleCase } from 'util/strings';
+import InstanceGroupLabels from 'components/InstanceGroupLabels';
 import CredentialChip from '../CredentialChip';
 import ChipGroup from '../ChipGroup';
 import { DetailList, Detail, UserDateDetail } from '../DetailList';
@@ -227,21 +228,7 @@ function PromptDetail({
                   label={t`Instance Groups`}
                   rows={4}
                   value={
-                    <ChipGroup
-                      numChips={5}
-                      totalChips={overrides.instance_groups.length}
-                      ouiaId="prompt-instance-groups-chips"
-                    >
-                      {overrides.instance_groups.map((instance_group) => (
-                        <Chip
-                          key={instance_group.id}
-                          ouiaId={`instance-group-${instance_group.id}-chip`}
-                          isReadOnly
-                        >
-                          {instance_group.name}
-                        </Chip>
-                      ))}
-                    </ChipGroup>
+                    <InstanceGroupLabels labels={overrides.instance_groups} />
                   }
                 />
               )}

awx/ui/src/components/Schedule/ScheduleDetail/ScheduleDetail.js

@@ -10,6 +10,7 @@ import useRequest, { useDismissableError } from 'hooks/useRequest';
 import { JobTemplatesAPI, SchedulesAPI, WorkflowJobTemplatesAPI } from 'api';
 import { parseVariableField, jsonToYaml } from 'util/yaml';
 import { useConfig } from 'contexts/Config';
+import InstanceGroupLabels from 'components/InstanceGroupLabels';
 import parseRuleObj from '../shared/parseRuleObj';
 import FrequencyDetails from './FrequencyDetails';
 import AlertModal from '../../AlertModal';
@@ -27,11 +28,6 @@ import { VariablesDetail } from '../../CodeEditor';
 import { VERBOSITY } from '../../VerbositySelectField';
 import getHelpText from '../../../screens/Template/shared/JobTemplate.helptext';
 
-const buildLinkURL = (instance) =>
-  instance.is_container_group
-    ? '/instance_groups/container_group/'
-    : '/instance_groups/';
-
 const PromptDivider = styled(Divider)`
   margin-top: var(--pf-global--spacer--lg);
   margin-bottom: var(--pf-global--spacer--lg);
@@ -498,26 +494,7 @@ function ScheduleDetail({ hasDaysToKeepField, schedule, surveyConfig }) {
                 fullWidth
                 label={t`Instance Groups`}
                 value={
-                  <ChipGroup
-                    numChips={5}
-                    totalChips={instanceGroups.length}
-                    ouiaId="instance-group-chips"
-                  >
-                    {instanceGroups.map((ig) => (
-                      <Link
-                        to={`${buildLinkURL(ig)}${ig.id}/details`}
-                        key={ig.id}
-                      >
-                        <Chip
-                          key={ig.id}
-                          ouiaId={`instance-group-${ig.id}-chip`}
-                          isReadOnly
-                        >
-                          {ig.name}
-                        </Chip>
-                      </Link>
-                    ))}
-                  </ChipGroup>
+                  <InstanceGroupLabels labels={instanceGroups} isLinkable />
                 }
                 isEmpty={instanceGroups.length === 0}
               />

awx/ui/src/components/WorkflowOutputNavigation/WorkflowOutputNavigation.js

@@ -24,12 +24,10 @@ function WorkflowOutputNavigation({ relatedJobs, parentRef }) {
   const { id } = useParams();
 
   const relevantResults = relatedJobs.filter(
-    ({
-      job: jobId,
-      summary_fields: {
-        unified_job_template: { unified_job_type },
-      },
-    }) => jobId && `${jobId}` !== id && unified_job_type !== 'workflow_approval'
+    ({ job: jobId, summary_fields }) =>
+      jobId &&
+      `${jobId}` !== id &&
+      summary_fields.job.type !== 'workflow_approval'
   );
 
   const [isOpen, setIsOpen] = useState(false);
@@ -101,16 +99,14 @@ function WorkflowOutputNavigation({ relatedJobs, parentRef }) {
           {sortedJobs?.map((node) => (
             <SelectOption
               key={node.id}
-              to={`/jobs/${
-                JOB_URL_SEGMENT_MAP[
-                  node.summary_fields.unified_job_template.unified_job_type
-                ]
-              }/${node.summary_fields.job?.id}/output`}
+              to={`/jobs/${JOB_URL_SEGMENT_MAP[node.summary_fields.job.type]}/${
+                node.summary_fields.job?.id
+              }/output`}
               component={Link}
-              value={node.summary_fields.unified_job_template.name}
+              value={node.summary_fields.job.name}
             >
               {stringIsUUID(node.identifier)
-                ? node.summary_fields.unified_job_template.name
+                ? node.summary_fields.job.name
                 : node.identifier}
             </SelectOption>
           ))}

awx/ui/src/components/WorkflowOutputNavigation/WorkflowOutputNavigation.test.js

@@ -0,0 +1,85 @@
+import React from 'react';
+import { within, render, screen, waitFor } from '@testing-library/react';
+import userEvent from '@testing-library/user-event';
+import WorkflowOutputNavigation from './WorkflowOutputNavigation';
+import { createMemoryHistory } from 'history';
+import { I18nProvider } from '@lingui/react';
+import { i18n } from '@lingui/core';
+import { en } from 'make-plural/plurals';
+import english from '../../../src/locales/en/messages';
+import { Router } from 'react-router-dom';
+
+jest.mock('react-router-dom', () => ({
+  ...jest.requireActual('react-router-dom'),
+  useParams: () => ({
+    id: 1,
+  }),
+}));
+const jobs = [
+  {
+    id: 1,
+    summary_fields: {
+      job: {
+        name: 'Ansible',
+        type: 'project_update',
+        id: 1,
+        status: 'successful',
+      },
+    },
+    job: 4,
+  },
+  {
+    id: 2,
+    summary_fields: {
+      job: {
+        name: 'Durham',
+        type: 'job',
+        id: 2,
+        status: 'successful',
+      },
+    },
+    job: 3,
+  },
+  {
+    id: 3,
+    summary_fields: {
+      job: {
+        name: 'Red hat',
+        type: 'job',
+        id: 3,
+        status: 'successful',
+      },
+    },
+    job: 2,
+  },
+];
+
+describe('<WorkflowOuputNavigation/>', () => {
+  test('Should open modal and deprovision node', async () => {
+    i18n.loadLocaleData({ en: { plurals: en } });
+    i18n.load({ en: english });
+    i18n.activate('en');
+    const user = userEvent.setup();
+    const ref = jest
+      .spyOn(React, 'useRef')
+      .mockReturnValueOnce({ current: 'div' });
+    const history = createMemoryHistory({
+      initialEntries: ['jobs/playbook/2/output'],
+    });
+    render(
+      <I18nProvider i18n={i18n}>
+        <Router history={history}>
+          <WorkflowOutputNavigation relatedJobs={jobs} parentRef={ref} />
+        </Router>
+      </I18nProvider>
+    );
+
+    const button = screen.getByRole('button');
+    await user.click(button);
+
+    await waitFor(() => screen.getByText('Workflow Nodes'));
+    await waitFor(() => screen.getByText('Red hat'));
+    await waitFor(() => screen.getByText('Durham'));
+    await waitFor(() => screen.getByText('Ansible'));
+  });
+});