Search: index in small chunks (#10914)

* Search: index in small chunks

The default is 500 documents, our documents have nested objects,
that can make the final document really big, and timeout ES.

Ref #10911 (comment)

* Test with 100

Author: stsewd

readthedocs/projects/tasks/search.py

@@ -215,6 +215,9 @@ def _create_imported_files_and_search_index(
             document=PageDocument,
             objects=html_files_to_index,
             index_name=search_index_name,
+            # Pages are indexed in small chunks to avoid a
+            # large payload that will probably timeout ES.
+            chunk_size=100,
         )
 
     # Remove old HTMLFiles from ElasticSearch

readthedocs/search/utils.py

@@ -10,7 +10,7 @@
 log = structlog.get_logger(__name__)
 
 
-def index_objects(document, objects, index_name=None):
+def index_objects(document, objects, index_name=None, chunk_size=500):
     if not DEDConfig.autosync_enabled():
         log.info("Autosync disabled, skipping searh indexing.")
         return
@@ -21,7 +21,7 @@ def index_objects(document, objects, index_name=None):
     if index_name:
         document._index._name = index_name
 
-    document().update(objects)
+    document().update(objects, chunk_size=chunk_size)
 
     # Restore the old index name.
     if index_name: