Merge pull request #25118 from edx/ormsbee/exempt-ccx-from-search-indexing

Exclude CCX Courses from search indexing.

cms/djangoapps/contentstore/tasks.py

@@ -16,6 +16,7 @@ from celery import group
 from celery.task import task
 from celery.utils.log import get_task_logger
 from celery_utils.persist_on_failure import LoggedPersistOnFailureTask
+from ccx_keys.locator import CCXLocator
 from django.conf import settings
 from django.contrib.auth import get_user_model
 from django.contrib.auth.models import User
@@ -180,6 +181,17 @@ def update_search_index(course_id, triggered_time_isoformat):
     """ Updates course search index. """
     try:
         course_key = CourseKey.from_string(course_id)
+
+        # We skip search indexing for CCX courses because there is currently
+        # some issue around Modulestore caching that makes it prohibitively
+        # expensive (sometimes hours-long for really complex courses).
+        if isinstance(course_key, CCXLocator):
+            LOGGER.warning(
+                u'Search indexing skipped for CCX Course %s (this is currently too slow to run in production)',
+                course_id
+            )
+            return
+
         CoursewareSearchIndexer.index(modulestore(), course_key, triggered_at=(_parse_time(triggered_time_isoformat)))
 
     except SearchIndexingError as exc:

cms/djangoapps/contentstore/tests/test_courseware_index.py

@@ -27,6 +27,7 @@ from contentstore.courseware_index import (
     SearchIndexingError
 )
 from contentstore.signals.handlers import listen_for_course_publish, listen_for_library_update
+from contentstore.tasks import update_search_index
 from contentstore.tests.utils import CourseTestCase
 from contentstore.utils import reverse_course_url, reverse_usage_url
 from course_modes.models import CourseMode
@@ -755,6 +756,20 @@ class TestTaskExecution(SharedModuleStoreTestCase):
         response = searcher.search(field_dictionary={"library": library_search_key})
         self.assertEqual(response["total"], 2)
 
+    def test_ignore_ccx(self):
+        """Test that we ignore CCX courses (it's too slow now)."""
+        # We're relying on our CCX short circuit to just stop execution as soon
+        # as it encounters a CCX key. If that isn't working properly, it will
+        # fall through to the normal indexing and raise an exception because
+        # there is no data or backing course behind the course key.
+        with patch('contentstore.courseware_index.CoursewareSearchIndexer.index') as mock_index:
+            self.assertIsNone(
+                update_search_index(
+                    "ccx-v1:OpenEdX+FAKECOURSE+FAKERUN+ccx@1", "2020-09-28T16:41:57.150796"
+                )
+            )
+            self.assertFalse(mock_index.called)
+
 
 @ddt.ddt
 class TestLibrarySearchIndexer(MixedWithOptionsTestCase):