Improve solr reindex performance by using a tuple instead a set for idxs

opengever/maintenance/scripts/repository_migration.py

@@ -1369,7 +1369,7 @@ def rename(self, items):
             # We do not need to reindex path as this seems to already happen
             # recursively
             self.add_to_reindexing_queue(
-                item['uid'], ('Title', 'sortable_title'))
+                item['uid'], ('Title', 'title_de', 'title_fr', 'title_en', 'sortable_title'))
             if not self.dry_run:
                 transaction.commit()
 
@@ -1433,7 +1433,10 @@ def reindex(self):
             if not obj:
                 logger.error("Could not find {} to reindex. Skipping".format(uid))
                 continue
-            obj.reindexObject(idxs=idxs)
+
+            # WARNING! idxs needs to be a tuple, otherwise solr will always update all attributes
+            # See: https://github.com/plone/collective.indexing/blob/2.0/src/collective/indexing/queue.py#L142
+            obj.reindexObject(idxs=tuple(idxs))
             if obj.portal_type == 'opengever.task.task':
                 # make sure that the model is up to date.
                 TaskSqlSyncer(obj, None).sync()