Merge pull request #5239 from rtfd/fix-reindexing

Support passing an explicit `index_name` for search indexing

Author: ericholscher

readthedocs/search/management/commands/reindex_elasticsearch.py

@@ -18,7 +18,7 @@
 class Command(BaseCommand):
 
     @staticmethod
-    def _get_indexing_tasks(app_label, model_name, queryset, document_class):
+    def _get_indexing_tasks(app_label, model_name, index_name, queryset, document_class):
         total = queryset.count()
         chunks = get_chunk(total, settings.ES_TASK_CHUNK_SIZE)
 
@@ -27,6 +27,7 @@ def _get_indexing_tasks(app_label, model_name, queryset, document_class):
                 'app_label': app_label,
                 'model_name': model_name,
                 'document_class': document_class,
+                'index_name': index_name,
                 'chunk': chunk
             }
             yield index_objects_to_es.si(**data)
@@ -62,6 +63,7 @@ def _run_reindex_tasks(self, models, queue):
 
             indexing_tasks = self._get_indexing_tasks(app_label=app_label, model_name=model_name,
                                                       queryset=queryset,
+                                                      index_name=new_index_name,
                                                       document_class=str(doc))
 
             post_index_task = switch_es_index.si(app_label=app_label, model_name=model_name,

readthedocs/search/tasks.py

@@ -10,7 +10,9 @@
 
 
 @app.task(queue='web')
-def index_objects_to_es(app_label, model_name, document_class, chunk=None, objects_id=None):
+def index_objects_to_es(
+    app_label, model_name, document_class, index_name=None, chunk=None, objects_id=None
+):
 
     assert not (chunk and objects_id), "You can not pass both chunk and objects_id"
 
@@ -31,9 +33,20 @@ def index_objects_to_es(app_label, model_name, document_class, chunk=None, objec
     elif objects_id:
         queryset = queryset.filter(id__in=objects_id)
 
+    if index_name:
+        # Hack the index name temporarily for reindexing tasks
+        old_index_name = document._doc_type.index
+        document._doc_type.index = index_name
+        log.info('Replacing index name %s with %s', old_index_name, index_name)
+
     log.info("Indexing model: {}, '{}' objects".format(model.__name__, queryset.count()))
     doc_obj.update(queryset.iterator())
 
+    if index_name:
+        log.info('Undoing index replacement, settings %s with %s',
+                 document._doc_type.index, old_index_name)
+        document._doc_type.index = old_index_name
+
 
 @app.task(queue='web')
 def delete_objects_in_es(app_label, model_name, document_class, objects_id):