More consistently provide fields in job_events logger

* Change scheme from using event dict to JobEvent object
* Add processing to grok object fields
* Allow override of provided formatter in case of future issues

awx/main/models/jobs.py

@@ -1289,10 +1289,10 @@ class JobEvent(CreatedModifiedModel):
         if event_data:
             artifact_dict = event_data.pop('artifact_data', None)
 
-        analytics_logger.info('Job event data saved.', extra=dict(event_model_data=kwargs))
-
         job_event = JobEvent.objects.create(**kwargs)
 
+        analytics_logger.info('Job event data saved.', extra=dict(python_objects=dict(job_event=job_event)))
+
         # Save artifact data to parent job (if provided).
         if artifact_dict:
             if event_data and isinstance(event_data, dict):

awx/main/tests/unit/utils/test_formatters.py

@@ -0,0 +1,24 @@
+from awx.main.models import Job, JobEvent
+
+from awx.main.utils.formatters import LogstashFormatter
+
+
+def test_log_from_job_event_object():
+    job = Job(id=4)
+    event = JobEvent(job_id=job.id)
+    formatter = LogstashFormatter()
+
+    data_for_log = formatter.reformat_data_for_log(
+        dict(python_objects=dict(job_event=event)), kind='job_events')
+
+    # Check entire body of data for any exceptions from getattr on event object
+    for fd in data_for_log:
+        if not isinstance(data_for_log[fd], basestring):
+            continue
+        assert 'Exception' not in data_for_log[fd], 'Exception delivered in data: {}'.format(data_for_log[fd])
+
+    # Verify existence of certain high-importance fields
+    for fd in ['changed', 'uuid', 'start_line', 'end_line', 'id', 'counter', 'host_name', 'stdout']:
+        assert fd in data_for_log
+
+    assert data_for_log['job'] == 4

awx/main/utils/formatters.py

@@ -37,13 +37,7 @@ class LogstashFormatter(LogstashFormatterVersion1):
         '''
         if kind == 'activity_stream':
             return raw_data
-        rename_fields = set((
-            'args', 'asctime', 'created', 'exc_info', 'exc_text', 'filename',
-            'funcName', 'id', 'levelname', 'levelno', 'lineno', 'module',
-            'msecs', 'msecs', 'message', 'msg', 'name', 'pathname', 'process',
-            'processName', 'relativeCreated', 'thread', 'threadName', 'extra',
-            'auth_token', 'tags', 'host', 'host_id', 'level', 'port', 'uuid'))
-        if kind == 'system_tracking':
+        elif kind == 'system_tracking':
             data = copy(raw_data['ansible_facts'])
         elif kind == 'job_events':
             data = copy(raw_data['event_model_data'])
@@ -51,7 +45,6 @@ class LogstashFormatter(LogstashFormatterVersion1):
             data = copy(raw_data)
         if isinstance(data, basestring):
             data = json.loads(data)
-        skip_fields = ('res', 'password', 'event_data', 'stdout')
         data_for_log = {}
 
         def index_by_name(alist):
@@ -86,18 +79,31 @@ class LogstashFormatter(LogstashFormatterVersion1):
                 return val
 
         if kind == 'job_events':
-            data.update(data.get('event_data', {}))
-            for fd in data:
-                if fd in skip_fields:
+            job_event = raw_data['python_objects']['job_event']
+            for field_object in job_event._meta.fields:
+
+                if not field_object.__class__ or not field_object.__class__.__name__:
+                    field_class_name = ''
+                else:
+                    field_class_name = field_object.__class__.__name__
+                if field_class_name in ['ManyToOneRel', 'ManyToManyField']:
                     continue
+
+                fd = field_object.name
                 key = fd
-                if fd in rename_fields:
-                    key = 'event_%s' % fd
-                val = data[fd]
-                if key.endswith('created'):
-                    time_float = time.mktime(data[fd].timetuple())
-                    val = self.format_timestamp(time_float)
-                data_for_log[key] = val
+                if field_class_name == 'ForeignKey':
+                    fd = '{}_id'.format(field_object.name)
+
+                try:
+                    data_for_log[key] = getattr(job_event, fd)
+                    if fd in ['created', 'modified'] and data_for_log[key] is not None:
+                        time_float = time.mktime(data_for_log[key].timetuple())
+                        data_for_log[key] = self.format_timestamp(time_float)
+                except Exception as e:
+                    data_for_log[key] = 'Exception `{}` producing field'.format(e)
+
+            data_for_log['event_display'] = job_event.get_event_display2()
+
         elif kind == 'system_tracking':
             data.pop('ansible_python_version', None)
             if 'ansible_python' in data:

awx/main/utils/handlers.py

@@ -26,6 +26,7 @@ from awx.main.utils.formatters import LogstashFormatter
 __all__ = ['HTTPSNullHandler', 'BaseHTTPSHandler', 'TCPHandler', 'UDPHandler',
            'configure_external_logger']
 
+
 logger = logging.getLogger('awx.main.utils.handlers')
 
 # AWX external logging handler, generally designed to be used
@@ -346,7 +347,15 @@ def configure_external_logger(settings_module, is_startup=True):
     if is_enabled:
         handler_class = HANDLER_MAPPING[settings_module.LOG_AGGREGATOR_PROTOCOL]
         instance = handler_class.from_django_settings(settings_module)
-        instance.setFormatter(LogstashFormatter(settings_module=settings_module))
+
+        # Obtain the Formatter class from settings to maintain customizations
+        configurator = logging.config.DictConfigurator(settings_module.LOGGING)
+        formatter_config = settings_module.LOGGING['formatters']['json'].copy()
+        formatter_config['settings_module'] = settings_module
+        formatter = configurator.configure_custom(formatter_config)
+
+        instance.setFormatter(formatter)
+
     awx_logger_instance = instance
     if is_enabled and 'awx' not in settings_module.LOG_AGGREGATOR_LOGGERS:
         awx_logger_instance = None

tools/docker-compose/README

@@ -1,2 +1,32 @@
 docker build --no-cache=true --rm=true -t ansible/tower_devel:latest .
 docker run --name tower_test -it --memory="4g" --cpuset="0,1" -v /Users/meyers/ansible/:/tower_devel -p 8013:8013 -p 8080:8080 -p 27017:27017 -p 2222:22 ansible/tower_devel
+
+## How to use the logstash container
+
+POST the following content to `/api/v1/settings/logging/` (this uses
+authentication set up inside of the logstash configuration file).
+
+```
+{
+    "LOG_AGGREGATOR_HOST": "logstash",
+    "LOG_AGGREGATOR_PORT": 8085,
+    "LOG_AGGREGATOR_TYPE": "logstash",
+    "LOG_AGGREGATOR_USERNAME": "awx_logger",
+    "LOG_AGGREGATOR_PASSWORD": "workflows",
+    "LOG_AGGREGATOR_LOGGERS": [
+        "awx",
+        "activity_stream",
+        "job_events",
+        "system_tracking"
+    ],
+    "LOG_AGGREGATOR_INDIVIDUAL_FACTS": false,
+    "LOG_AGGREGATOR_TOWER_UUID": "991ac7e9-6d68-48c8-bbde-7ca1096653c6",
+    "LOG_AGGREGATOR_ENABLED": true
+}
+```
+
+An example of how to view the most recent logs from the container:
+
+```
+docker exec -i -t $(docker ps -aqf "name=tools_logstash_1") tail -n 50 /logstash.log
+```