Allow limiting the ansible event datastructure

This adds two tunables:

* omit_event_data: excludes everything under the event['event_data']
                   property
* only_failed_event_data: As above, but will include event data for
                          failed events

The idea here is to allow the caller to further constrain the size of
the data payload returned to the interested system.

Author: matburt

ansible_runner/__main__.py

@@ -307,6 +307,22 @@ def main(sys_args=None):
              "Ansible output (default=False)"
     )
 
+    runner_group.add_argument(
+        "--omit-event-data",
+        action="store_true",
+        help="Omits including extra event data in the callback payloads "
+             "or the Runner payload data files "
+             "(status and stdout still included)"
+    )
+
+    runner_group.add_argument(
+        "--only-failed-event-data",
+        action="store_true",
+        help="Only adds extra event data for failed tasks in the callback "
+             "payloads or the Runner payload data files "
+             "(status and stdout still included for other events)"
+    )
+
     runner_group.add_argument(
         "-q", "--quiet",
         action="store_true",
@@ -515,6 +531,8 @@ def main(sys_args=None):
                                    rotate_artifacts=args.rotate_artifacts,
                                    ignore_logging=False,
                                    json_mode=args.json,
+                                   omit_event_data=args.omit_event_data,
+                                   only_failed_event_data=args.only_failed_event_data,
                                    inventory=args.inventory,
                                    forks=args.forks,
                                    project_dir=args.project_dir,

ansible_runner/display_callback/events.py

@@ -134,40 +134,45 @@ def get(self):
         return ctx
 
     def get_begin_dict(self):
+        omit_event_data = os.getenv("RUNNER_OMIT_EVENTS", "False").lower() == "true"
+        include_only_failed_event_data = os.getenv("RUNNER_ONLY_FAILED_EVENTS", "False").lower() == "true"
         event_data = self.get()
-        # TODO: Need to rework these values
+        event = event_data.pop('event', None)
+        if not event:
+            event = 'verbose'
+            for key in ('debug', 'verbose', 'deprecated', 'warning', 'system_warning', 'error'):
+                if event_data.get(key, False):
+                    event = key
+                    break
+        event_dict = dict(event=event)
+        should_process_event_data = (include_only_failed_event_data and event in ('runner_on_failed', 'runner_on_async_failed', 'runner_on_item_failed')) \
+            or not include_only_failed_event_data
         if os.getenv('JOB_ID', ''):
-            event_data['job_id'] = int(os.getenv('JOB_ID', '0'))
+            event_dict['job_id'] = int(os.getenv('JOB_ID', '0'))
         if os.getenv('AD_HOC_COMMAND_ID', ''):
-            event_data['ad_hoc_command_id'] = int(os.getenv('AD_HOC_COMMAND_ID', '0'))
+            event_dict['ad_hoc_command_id'] = int(os.getenv('AD_HOC_COMMAND_ID', '0'))
         if os.getenv('PROJECT_UPDATE_ID', ''):
-            event_data['project_update_id'] = int(os.getenv('PROJECT_UPDATE_ID', '0'))
-        event_data.setdefault('pid', os.getpid())
-        event_data.setdefault('uuid', str(uuid.uuid4()))
-        event_data.setdefault('created', datetime.datetime.utcnow().isoformat())
+            event_dict['project_update_id'] = int(os.getenv('PROJECT_UPDATE_ID', '0'))
+        event_dict['pid'] = event_data.get('pid', os.getpid())
+        event_dict['uuid'] = event_data.get('uuid', str(uuid.uuid4()))
+        event_dict['created'] = event_data.get('created', datetime.datetime.utcnow().isoformat())
         if not event_data.get('parent_uuid', None):
             for key in ('task_uuid', 'play_uuid', 'playbook_uuid'):
                 parent_uuid = event_data.get(key, None)
                 if parent_uuid and parent_uuid != event_data.get('uuid', None):
-                    event_data['parent_uuid'] = parent_uuid
-                    break
-
-        event = event_data.pop('event', None)
-        if not event:
-            event = 'verbose'
-            for key in ('debug', 'verbose', 'deprecated', 'warning', 'system_warning', 'error'):
-                if event_data.get(key, False):
-                    event = key
+                    event_dict['parent_uuid'] = parent_uuid
                     break
-        max_res = int(os.getenv("MAX_EVENT_RES", 700000))
-        if event not in ('playbook_on_stats',) and "res" in event_data and len(str(event_data['res'])) > max_res:
-            event_data['res'] = {}
-        event_dict = dict(event=event, event_data=event_data)
-        for key in list(event_data.keys()):
-            if key in ('job_id', 'ad_hoc_command_id', 'project_update_id', 'uuid', 'parent_uuid', 'created',):
-                event_dict[key] = event_data.pop(key)
-            elif key in ('verbosity', 'pid'):
-                event_dict[key] = event_data[key]
+        else:
+            event_dict['parent_uuid'] = event_data.get('parent_uuid', None)
+        if "verbosity" in event_data.keys():
+            event_dict["verbosity"] = event_data.pop("verbosity")
+        if not omit_event_data and should_process_event_data:
+            max_res = int(os.getenv("MAX_EVENT_RES", 700000))
+            if event not in ('playbook_on_stats',) and "res" in event_data and len(str(event_data['res'])) > max_res:
+                event_data['res'] = {}
+        else:
+            event_data = dict()
+        event_dict['event_data'] = event_data
         return event_dict
 
     def get_end_dict(self):

ansible_runner/interface.py

@@ -124,6 +124,8 @@ def run(**kwargs):
     :param fact_cache: A string that will be used as the name for the subdirectory of the fact cache in artifacts directory.
                        This is only used for 'jsonfile' type fact caches.
     :param fact_cache_type: A string of the type of fact cache to use.  Defaults to 'jsonfile'.
+    :param omit_event_data: Omits extra ansible event data from event payload (stdout and event still included)
+    :param only_failed_event_data: Omits extra ansible event data unless it's a failed event (stdout and event still included)
     :type private_data_dir: str
     :type ident: str
     :type json_mode: bool
@@ -155,6 +157,8 @@ def run(**kwargs):
     :type directory_isolation_base_path: str
     :type fact_cache: str
     :type fact_cache_type: str
+    :type omit_event_data: bool
+    :type only_failed_event_data: bool
 
     :returns: A :py:class:`ansible_runner.runner.Runner` object
     '''

ansible_runner/runner_config.py

@@ -79,7 +79,8 @@ def __init__(self,
                  process_isolation=False, process_isolation_executable=None, process_isolation_path=None,
                  process_isolation_hide_paths=None, process_isolation_show_paths=None, process_isolation_ro_paths=None,
                  tags=None, skip_tags=None, fact_cache_type='jsonfile', fact_cache=None, project_dir=None,
-                 directory_isolation_base_path=None, envvars=None, forks=None, cmdline=None):
+                 directory_isolation_base_path=None, envvars=None, forks=None, cmdline=None, omit_event_data=False,
+                 only_failed_event_data=False):
         self.private_data_dir = os.path.abspath(private_data_dir)
         self.ident = str(ident)
         self.json_mode = json_mode
@@ -127,6 +128,8 @@ def __init__(self,
         self.envvars = envvars
         self.forks = forks
         self.cmdline_args = cmdline
+        self.omit_event_data = omit_event_data
+        self.only_failed_event_data = only_failed_event_data
 
     def prepare(self):
         """
@@ -200,6 +203,9 @@ def prepare(self):
             self.env['ANSIBLE_CACHE_PLUGIN'] = 'jsonfile'
             self.env['ANSIBLE_CACHE_PLUGIN_CONNECTION'] = self.fact_cache
 
+        self.env["RUNNER_OMIT_EVENTS"] = str(self.omit_event_data)
+        self.env["RUNNER_ONLY_FAILED_EVENTS"] = str(self.only_failed_event_data)
+
     def prepare_inventory(self):
         """
         Prepares the inventory default under ``private_data_dir`` if it's not overridden by the constructor.

test/integration/test_events.py

@@ -37,7 +37,10 @@ def test_basic_events(is_run_async=False,g_facts=False):
 
     okay_event = okay_events[0]
     assert "uuid" in okay_event and len(okay_event['uuid']) == 36
+    assert "parent_uuid" in okay_event and len(okay_event['parent_uuid']) == 36
     assert "stdout" in okay_event and len(okay_event['stdout']) > 0
+    assert "start_line" in okay_event and int(okay_event['start_line']) > 0
+    assert "end_line" in okay_event and int(okay_event['end_line']) > 0
     assert "event_data" in okay_event and len(okay_event['event_data']) > 0
 
 
@@ -54,6 +57,24 @@ def test_basic_serializeable():
     json.dumps(events)
 
 
+def test_event_omission():
+    tdir = tempfile.mkdtemp()
+    r = run(private_data_dir=tdir,
+            inventory="localhost ansible_connection=local",
+            omit_event_data=True,
+            playbook=[{'hosts': 'all', 'gather_facts': False, 'tasks': [{'debug': {'msg': "test"}}]}])
+    assert not any([x['event_data'] for x in r.events])
+
+
+def test_event_omission_except_failed():
+    tdir = tempfile.mkdtemp()
+    r = run(private_data_dir=tdir,
+            inventory="localhost ansible_connection=local",
+            only_failed_event_data=True,
+            playbook=[{'hosts': 'all', 'gather_facts': False, 'tasks': [{'fail': {'msg': "test"}}]}])
+    all_event_datas = [x['event_data'] for x in r.events if x['event_data']]
+    assert len(all_event_datas) == 1
+
 @pytest.mark.skipif(LooseVersion(pkg_resources.get_distribution('ansible').version) < LooseVersion('2.8'),
                     reason="Valid only on Ansible 2.8+")
 def test_runner_on_start(rc):