Check for app_context and request in g to prevent Attribute Errors.

We can add a request_id for tasks that are not spawned by an HTTP request, for example scheduled or nightly tasks. That means you can match up all the tasks spawned by a single task, for example, create-night-billing spawns 4 tasks, those would all have the same idea. Not sure if that is helpful or not. Also it might be confusing to have a request_id for logs that were not started from a request so I have left it out.

app/celery/celery.py

@@ -1,10 +1,9 @@
 import time
-
 from gds_metrics.metrics import Histogram
 from celery import Celery, Task
 from celery.signals import worker_process_shutdown
 from flask import g, request
-from flask.ctx import has_request_context
+from flask.ctx import has_request_context, has_app_context
 
 
 @worker_process_shutdown.connect
@@ -55,10 +54,9 @@ def make_task(app):
         def apply_async(self, args=None, kwargs=None, task_id=None, producer=None,
                         link=None, link_error=None, **options):
             kwargs = kwargs or {}
-
             if has_request_context() and hasattr(request, 'request_id'):
                 kwargs['request_id'] = request.request_id
-            elif g.request_id:
+            elif has_app_context() and 'request_id' in g:
                 kwargs['request_id'] = g.request_id
 
             with SQS_APPLY_ASYNC_DURATION_SECONDS.labels(self.name).time():